diff --git a/core/claims/DDD-address-02.yaml b/core/claims/DDD-address-02.yaml index 6fd5753..b8a95f3 100644 --- a/core/claims/DDD-address-02.yaml +++ b/core/claims/DDD-address-02.yaml @@ -21,3 +21,10 @@ notes: | Emil's ruling R-6, 2026-09-27, in his words: "our choice to reuse the same act in the code is an optimization not the truth." Filed as a normative claim: it is a governing choice about representation, and its test is utility. + + Appended 2026-11-18 (DDD-dec-122; G1-Q10, option 1, in + meta/sessions/2026-11-18-act-map-scope-and-pull/rulings.md). The region's "Governs the act map + (a tree of occurrences)" is superseded by DDD-address-06: the act map's nodes are act sites, + the tree is containment under each top act, and occurrences are recorded beside the map. The + region is not reworded. This claim's statement, falsifier, test and breaks stand: every + occurrence is still its own act. diff --git a/core/claims/DDD-address-06.yaml b/core/claims/DDD-address-06.yaml new file mode 100644 index 0000000..fefbfdd --- /dev/null +++ b/core/claims/DDD-address-06.yaml @@ -0,0 +1,78 @@ +format: 1 +id: DDD-address-06 +kind: conceptual +statement: | + The act map is the record of one ruled decomposition of the work inside one declared scope into acts, read under one declaration of ground axes, and it holds the act sites at their addresses, the composite acts that layer them, the causal dependencies between them and the index of the determinables that make demand arrive in its scope. +status: projected +region: | + Diachronic — it relates acts across occurrences, so it files here and not upstream + (DDD-dec-09's charter). One map per scope and per declaration of ground axes (DDD-address-07). + Its parts: + + - Scope. The map's first declaration: the decisions its holders take. Everything outside it + is world to the map, including other people's decisions. A decision outside reaches the map + only as ground crossing its edge (2026-11-18 R-2). + - Act sites. One site per address (DDD-ground-06): a ground position on the declared axes and + a layer path, with the occupant left out. Occurrences are not in the map. Each occurrence is + recorded against exactly one site, with its time index (DDD-address-01), in a record beside + the map. + - Layers. Composite acts above an act and below it, and no other kind of layer + (DDD-ground-06; upstream core/09 §1). Containment is a tree under each top act, and a top + act is a composite act with none above it. A map may hold several top acts. + - Dependencies. Site B depends on site A when a state change an occurrence at A produced is + in the ground reading of an occurrence at B (DDD-address-01). A directed graph, not a tree. + It is acyclic over occurrences. Over sites it may loop. + - The index of determinables. One row per determinable at or inside the edge: its outcomes, + its entry act, the act where its determinate lands, its specified, observed and expected + states (DDD-address-11; actual is never read), its kind as a position on the map's declared + kind axis (DDD-address-10) and its rate of change. The observed state is a pointer to the + latest observation in the occurrence record, not a copy, so the map stays structure and the + record holds what changes. A row whose specified state cites a decision in force has been + taken up by governance inside the scope, which makes it a decision anchored to acts + (DDD-ground-01). + + The map has no direction of its own beyond causal order. A view gives it one by naming a + north (DDD-address-08). It is central as an index every later artefact attaches to, not as a + container that holds them: each of method steps 2 to 10 attaches to a part of it or to a + record beside it (DDD-dec-122, S-1). +evidence: [] +falsifier: | + An act site that must sit within two composite acts for its governing set to be stated, a dependency that can be stated between two occurrences and between no two sites, a determinable that makes demand arrive in the scope and can be given no entry act on the map, or a composite act that cannot be placed wholly inside or wholly outside the declared scope. +test: | + T2b, the first independent application case: its act map is built in the five parts, and each falsifier limb is searched for (registered again as T-1 in meta/sessions/2026-11-18-act-map-scope-and-pull/effects.md). The third limb also yields a finding when it fires — work the decomposition does not cover. +breaks: | + First limb: the layer path, and DDD-ground-06's exactly-one-address reading. Second limb: the split between the map and the occurrence record, and with it the lifecycle graph data (applications/pdlc/graph/, applications/sdlc/acts/). Third limb: that the map shows where demand comes from at all, and DDD-address-08 and -09 with it. Fourth limb: scope as a declaration made before any act is mapped, and the first part of step 1. +owner: none +changed: 2026-11-18, act map, scope and pull session (G2, on the G1 rulings; DDD-dec-122) +notes: | + History. First drafted on 2026-11-17 on a branch that did not land + (claude/act-map-and-process-2026-11-17; its commits are kept on claude/pull-and-scope-2026-11-18 + at 47cf4cc), and amended on the rulings of 2026-11-18 (meta/sessions/2026-11-18-pull-and-scope/ + rulings.md, R-8). The amendment adds the scope, the readable states and kind in the index, the + observed pointer and the fourth falsifier limb. + + Ruled by Emil: 2026-11-17 E-2, R-3, R-6 and R-7 points 1 and 2 + (meta/sessions/2026-11-17-act-map-and-process/rulings.md); 2026-11-18 R-1, R-2, R-4 and R-5; + G1-Q1 and G1-Q8 of meta/sessions/2026-11-18-act-map-scope-and-pull/rulings.md. + + Emil on the layers (2026-11-17 E-2): "We have the layering its the acts at ground axes and then + we have composite acts below and above that - it can never be anything else." On scope + (2026-11-18 R-2): "scope must be the first else we cant place the pull, or we would end up far + far from our area of influence". + + Prose that contradicts this claim, flagged and not harmonised: + - DDD-address-02's region calls the act map "a tree of occurrences". By G1-Q10, option 1, a + note appended to DDD-address-02 records that the phrase is superseded by this claim. The + region is not reworded. + - applications/sdlc/sdlc.md §10 says a second top act's acts "sit at the same addresses with a + different layer path". The layer path is part of the address (DDD-ground-06), so they can + share a ground position (DDD-address-04) and not an address. Open, successor item O-2 of + 2026-11-17. + - Four places name the SDLC's part of the lifecycle map "the act map" (O-3 of 2026-11-17). + + UNVERIFIED — Emil review: + - "Top act" is defined here as a position in the layering, not as a new kind. + - That occurrences sit outside the map rests on the lc: vocabulary (lc:Act, "an act site of + the map: one address, possibly many occurrences"; Round 28, R30) and on no other claim. + - The index has not been built once. Building the lifecycle's index is out of #80's scope. + - The rate of change has no unit and no measure (O-7 of 2026-11-17). diff --git a/core/claims/DDD-address-07.yaml b/core/claims/DDD-address-07.yaml new file mode 100644 index 0000000..9cdf338 --- /dev/null +++ b/core/claims/DDD-address-07.yaml @@ -0,0 +1,64 @@ +format: 1 +id: DDD-address-07 +kind: normative +statement: | + A second act map exists when, and only when, a second declaration exists: a second scope, or a second declaration of ground axes. +status: projected +region: | + Diachronic — a relation between records of acts (downstream per DDD-dec-09's charter). Scope + says what is world. Axes say what is comparable (G1-Q2). Two act sites belong to one map + exactly when they are inside one declared scope and their positions are read on the same + declared axes, under one ruled decomposition (DDD-ground-06's region). It follows that: + + - a further top act under the same scope and axes is a further tree in the same map; + - a part of a map, such as the acts within one composite act, is not a map; + - a view of a map (DDD-address-08) is not a map; + - the same work decomposed a second way gives one map at a time, because re-decomposition + supersedes (DDD-ground-06's region); + - a wider map may declare a narrower scope within it. A pull of the narrower scope then shows + in the wider map as a seam at that scope's edge. Where the wider map does not declare the + narrower scope, the pull shows there as a dependency only (2026-11-18 R-3). + + The lifecycle (applications/pdlc/pdlc.md §11) is one map, with the PDLC as its scope and the + How declared within it. A map with the How alone as its scope would be a second map on the + same axes. + + Between maps with different declarations of ground axes, coincidence (DDD-address-04) and + applicability predicates (DDD-ground-01) are not defined. One relation remains: an act in one + reads state that an act in the other produced. +evidence: + - kind: observation + ref: "meta/sessions/2026-11-17-act-map-and-process/survey.md, findings S-6 and S-7" + note: | + Three mapped cases in the repository, each with its own declaration of axes: the dry run's + T2a (four axes), the quick start's Varve slice (three) and the product lifecycle (three, + DDD-dec-115). No map uses two declarations and no declaration serves two maps. The accepted + T2a map (DDD-dec-38) has three top acts under one declaration. An observation of how the + word has been used, written inside the framework: it does not test the rule. +falsifier: | + Between maps with different declarations of ground axes, a governing decision whose applicability predicate must cover occurrences in both, or two occurrences in the two maps whose behaviour must be compared to answer a question the work poses; between maps of nested scopes on the same axes, shared acts that carry different addresses in the two maps. +test: | + Utility, on the second application case: an organisation mapped beside the PDLC either shares the PDLC's scope and axes or does not, and the rule holds if no governing decision, no behaviour comparison and no shared act has to cross the boundary it draws in the way the falsifier names. +breaks: | + Checking a map on its own (applications/pdlc/graph/check.py), the two grains of DDD-address-05, which assume one address space, and any reading of one map that is then given two scopes. +owner: none +changed: 2026-11-18, act map, scope and pull session (G2, on the G1 rulings; DDD-dec-122) +notes: | + History. First drafted on 2026-11-17 on a branch that did not land + (claude/act-map-and-process-2026-11-17; kept on claude/pull-and-scope-2026-11-18 at 47cf4cc), + and amended on the rulings of 2026-11-18 (meta/sessions/2026-11-18-pull-and-scope/rulings.md, + R-8). The 2026-11-17 statement read "A second act map exists when, and only when, a second + declaration of ground axes exists" (2026-11-17 R-7 point 3). G1-Q2 supersedes that wording, + adding the scope (meta/sessions/2026-11-18-act-map-scope-and-pull/rulings.md). + + Correction by Emil (C-1): the falsifier first proposed kept the 2026-11-17 limb for every pair + of maps. "As written, any PDLC decision covering an SDLC act fires it on day one." The limb now + holds only between different declarations of axes, and nested scopes have their own. + + Filed as a normative claim, as DDD-address-02 is: a governing choice about representation, + tested by utility. + + UNVERIFIED — Emil review: whether two pieces of work share a scope or a declaration is decided + by whoever rules the decomposition. Any two maps on the same axes can be made one by declaring + the union of their scopes. The rule says what a second map is and what it costs. It does not + say when to make one. diff --git a/core/claims/DDD-address-08.yaml b/core/claims/DDD-address-08.yaml new file mode 100644 index 0000000..54ece8e --- /dev/null +++ b/core/claims/DDD-address-08.yaml @@ -0,0 +1,78 @@ +format: 1 +id: DDD-address-08 +kind: normative +statement: | + A process is a human understanding of work, scoped within an area of expertise and not an object of the model, and its translation into the model is a named view of one act map, fixed by one or more norths and optionally bounded by composite acts inside the map's scope, where a north is the determinable of a pull whether or not a gap is open on it now. +status: projected +region: | + Wherever work that its holders call a process is mapped with the method. The view shows the + act sites on a dependency path (DDD-address-01) from each north's entry act to the act where + its determinate lands, both inside the map's scope (DDD-address-06), and inside the bound where + one is given. It leaves out readings of earlier occurrences, so rank from north is strict. Its + parts: + + - North. A determinable from the map's index: a cause, not a decision and not an act. It is + the determinable of a pull (DDD-address-09) whether or not a gap is open on it now, so a + process keeps its north between occurrences. The determinable is to the gap as the act site + is to the occurrence. It is a north whether or not governance inside the scope has taken it + up (DDD-frame-13; upstream core/00 §4a). A decision taken outside the scope is world to the + map. Time sets its rate of change, and that change is what creates demand, so work that + starts on a calendar has a north like any other. + - The other end. The act at which discharge produces the determinate (upstream core/09 §7a). + - The bound. None, one or more composite acts inside the scope (DDD-ground-06). People name a + process by where the work happens and who holds it as well as by why it happens, and the + model keeps those apart: the first is composite acts, the second is norths. + + A process is a saved view with a name. It is never a node of the map, and no decision is + anchored to it. A view is not a second act map (DDD-address-07). The translation may cross + composite acts. +evidence: [] +falsifier: | + A view that the holders of a process do not recognise as their process, or a process whose holders can name no determinable it answers once the calendar is set aside, or a process its holders recognise whose work answers no gap between any two states at any time. +test: | + Utility, on a second mapped case outside software: the holders of a process there are shown its view, and the claim fails on any limb. A process is settled inside one area of expertise, so the test is of use to its holders and not of fact. +breaks: | + Step 1b's use of "a process" as its unit (method/01b-map-the-flow-of-demand.md), the promotion condition of DDD-flow-01 to DDD-flow-04 ("a second process"), which would have no defined subject, and under the third limb the identification of north with a pull's determinable. +owner: none +changed: 2026-11-18, act map, scope and pull session (G2, on the G1 rulings; DDD-dec-122) +notes: | + History. First drafted on 2026-11-17 on a branch that did not land + (claude/act-map-and-process-2026-11-17; kept on claude/pull-and-scope-2026-11-18 at 47cf4cc), + and amended on the rulings of 2026-11-18 (meta/sessions/2026-11-18-pull-and-scope/rulings.md, + R-8). The 2026-11-17 statement fixed the view by "unresolved determinables". The amendment + makes north the determinable of a pull, bounds the view inside the scope, and drops + "unresolved" as a condition. + + Ruled by Emil: 2026-11-17 E-1, R-2, R-4, R-5, R-6 and R-7 point 4; 2026-11-18 R-2 and R-5; + G1-Q3 with correction C-2, and K-1 (meta/sessions/2026-11-18-act-map-scope-and-pull/ + rulings.md). + + K-1, ruled 2026-11-18 13:26: C-2 supersedes the wording of 2026-11-17 R-2 ("North is the + unresolved determinable") and of the matching line in issue #80's "ruled before" list ("North + is an unresolved determinable, where demand originates"), not their meaning. A north is a cause + and a determinable, as R-2 ruled. It rests on 2026-11-17 R-6: time sets a determinable's rate + of change, which is what creates demand. + + Emil on the process (2026-11-17 R-4): "A process is a simple human understanding og work being + done, it is scoped and settled usually in a way that only makes sense within a given area of + expertise. Software, finance, hr etc. What we have is a much more complex model of the world + and mapping a process into our model is going to be hard. But I still think its the best + translation of the concept we have". On north (2026-11-17 E-4): "North cant be a decision it + must be where the demand for decisions originate, so it is a cause." + + Corrections, recorded in the session rulings with attribution: + - Claude first proposed "a process is a composite act". Withdrawn on 2026-11-17 E-2. + - 2026-11-17 R-1 ruled north to be the discharge act of a declared demand kind. Superseded by + R-2: north is the determinable, and the discharge act is the other end. + - Claude proposed north to be "a decision that must be made". Withdrawn on 2026-11-17 E-4. + - Claude kept calendar-started work as a candidate process with no north. Withdrawn on E-5. + - Claude made "unresolved" a condition of north at G1 (P-3). Corrected by Emil (C-2). + + What this leaves as filed: DDD-dec-114's Round 9 reading of the SDLC as a composite act within + the PDLC. In this claim's terms the SDLC is a bound, and in DDD-address-06's it may also be + declared a scope. The ruling made in Mantle's planning project on 2026-11-17, "a process is a + composite act that holds acts directly", is a view with a bound and no north. Whether Mantle + keeps it is ruled there (O-9 of 2026-11-17). + + UNVERIFIED — Emil review: that a process may cross composite acts without the decomposition + being wrong is Claude's reading of 2026-11-17 R-4. diff --git a/core/claims/DDD-address-09.yaml b/core/claims/DDD-address-09.yaml new file mode 100644 index 0000000..cd9c1d7 --- /dev/null +++ b/core/claims/DDD-address-09.yaml @@ -0,0 +1,57 @@ +format: 1 +id: DDD-address-09 +kind: conceptual +statement: | + A pull is a gap between two of the four states of a determinable (DDD-address-11), read inside a map's declared scope, where the determinable has two or more outcomes that cannot all hold together, and a pull is placed by a trace back from the request that stops where the answer lies outside the scope. +status: projected +region: | + Wherever a map's scope is declared (DDD-address-06). A pull is found at the edge first: what + crosses into the scope and out of it. Its entry act is the first act inside the scope that + reads it. Under a different scope the same determinable may show a different gap and a + different entry act. + + The gaps are examples, not an exhaustive list (G1-Q6): expected with nothing specified (a + need nobody has taken up); specified with nothing actual (an approved change not yet built); + specified against actual (a deviation, a bug); actual against observed (a deviation nobody has + seen, or an alarm with nothing behind it); expected against specified (the specification is + wrong); expected against observed (the software matches its specification, and the holder + still cannot do the thing). + + Actual is never read directly (DDD-address-11). "Specified against observed" is the readable + stand-in for "specified against actual", and only a stand-in. A deviation is found once any + holder observes an outcome outside their tolerance, whether or not they trace it to the + software: that is "expected against observed", found from the expected side (G1-W). Exposure + nobody has observed is latent, which upstream canon calls escaped (term:escape), and a pull + procedure does not find it. Finding it before it lands is the work of assurance, in method + step 5 (K-3). +evidence: [] +falsifier: | + A request inside a declared scope that its holders recognise as demand and that shows no gap between any two states of any determinable at or inside the edge, or two practitioners who, given the same scope and the same record, independently write statements for one pull that name different entry acts. +test: | + Reproducibility: two practitioners run the five-part procedure of step 1 on one scope without conferring, and their pull statements are compared field by field. The first case is T2b (registered again as T-2 and T-3 in meta/sessions/2026-11-18-act-map-scope-and-pull/effects.md). +breaks: | + Step 1's first part, which starts from scope and pull; DDD-address-08's north as the determinable of a pull; the pull statement and its checks. +owner: none +changed: 2026-11-18, act map, scope and pull session (G2, on the G1 rulings; DDD-dec-122) +notes: | + New on 2026-11-18. Ruled by Emil: 2026-11-18 R-1 ("this should be two or more outcomes"), R-2 + (meta/sessions/2026-11-18-pull-and-scope/rulings.md), and G1-Q5, G1-Q6, G1-W and K-3 + (meta/sessions/2026-11-18-act-map-scope-and-pull/rulings.md). + + Asked for by Emil at 2026-11-18 E-3: "we need a definition of it so that it can be reproduced + by practicitioners". The test is that reproduction. + + Weakest point, as ruled at K-3 to stay: the gap between actual and observed can never be + inspected (upstream term:outcome, "the world renders outcomes, never verdicts"). So a + deviation nobody has seen is a pull this claim names and its procedure cannot find. + + Emil's sentence, recorded in his words and not a rule (K-3): "a bug nobody has found will + never be found. and a bug that is never found has no impact." It holds of a deviation never + found by anyone. Nobody can know "never" in advance. Claude caught the conflict with + term:escape ("Latent defect exposure. The only forbidden state.") before filing. + + UNVERIFIED — Emil review: that a pull is found at the edge first is Claude's reading of + 2026-11-18 E-4 ("they also inspect a system and look at the input and output"). + + The procedure and its worked example (the acme shop, with the How as the scope): + meta/sessions/2026-11-18-act-map-scope-and-pull/pull-procedure.md. diff --git a/core/claims/DDD-address-10.yaml b/core/claims/DDD-address-10.yaml new file mode 100644 index 0000000..12c0712 --- /dev/null +++ b/core/claims/DDD-address-10.yaml @@ -0,0 +1,37 @@ +format: 1 +id: DDD-address-10 +kind: conceptual +statement: | + A kind of demand is a position on a declared ground axis, so kinds differ by the ground a demand pulls from and not by where it came from, and two demand items at one position are of one kind whatever their source. +status: projected +region: | + Every map (DDD-address-06). Each map declares the axis its kinds are read on (G1-Q4). The + method offers "built or not built" as the starting point, as it offers the PDLC's kinds: a + feature's ground is that it has not been built, and a bug's is that it has (2026-11-18 R-4). + A demand item's source is kept as where it was sighted, and decides nothing about its kind. + Consistent with DDD-flow-01, whose three PDLC kinds are each tested relative to the decisions + in force, which is a reading of ground. +evidence: [] +falsifier: | + Two demand items at one ground position that must be ordered by different owners to be ordered on the ground they need, or one item whose kind changes with who reported it while its ground position is unchanged. +test: | + Counterexample hunt over the demand items of each mapped case, starting with T2b (registered again as T-4): each item's kind is read from its position on the map's declared axis, and the falsifier's two limbs are searched for. +breaks: | + Step 1b's classification of each source of pull into kinds, and the tables that classify by source in method/01b-map-the-flow-of-demand.md and applications/pdlc/pdlc.md §9, read as classification by source. Not DDD-flow-01. +owner: none +changed: 2026-11-18, act map, scope and pull session (G2, on the G1 rulings; DDD-dec-122) +notes: | + New on 2026-11-18. Ruled by Emil: 2026-11-18 R-4 (meta/sessions/2026-11-18-pull-and-scope/ + rulings.md) and G1-Q4 and G1-Q5 (meta/sessions/2026-11-18-act-map-scope-and-pull/rulings.md). + + Emil (2026-11-18 R-4): "a bug and a feature cant share ground because ground for a feature is + by necessity that it hasnt been implemented and a bug must have been implemented"; "yes - it + must be else it cant be pull if we cant determine […] the ground it pulls from". + + [PROPOSED], not ruled (G1-Q4): a second axis, "whose determinable" (the product's or the + arrangement's), which places own-ground demand. Own-ground demand is its only evidence. On + "built or not built" alone, the acme shop's #104 (own-ground) sits with #103 (owed) + (meta/sessions/2026-11-18-act-map-scope-and-pull/pull-procedure.md, G1-A3). + + Claude's withdrawn proposal of 2026-11-18, an origin field on the pull statement, is recorded + in that day's rulings.md (corrections, on R-4). diff --git a/core/claims/DDD-address-11.yaml b/core/claims/DDD-address-11.yaml new file mode 100644 index 0000000..a8c91d5 --- /dev/null +++ b/core/claims/DDD-address-11.yaml @@ -0,0 +1,44 @@ +format: 1 +id: DDD-address-11 +kind: conceptual +statement: | + A determinable is read in four states inside a scope — specified, the governing decisions in force over it; actual, the outcome, the determinate as it lands in the world; observed, the verdict, the same determinate as assessed by a declared predicate; and expected, the tolerance held by each holder who bears the outcome — and actual is never read directly. +status: projected +region: | + Wherever a map's scope is declared (DDD-address-06). Specified, observed and expected are + readable. Observed is what someone inside the scope has seen. A sighting made outside reaches + the map across its edge as ground. Expected is not one state: each holder who bears the + outcome has one, and holders who disagree are all kept. Actual is never read, because the + world renders outcomes and not verdicts (upstream term:outcome), so a pull statement records + it as "not read". The gaps between the states are the pulls (DDD-address-09). +evidence: [] +falsifier: | + A demand item whose gap cannot be stated without a fifth state, or a determinable outside software for which one of the four cannot be given any meaning. +test: | + Counterexample hunt over the pull statements of each mapped case, starting with T2b (registered again as T-5): actual is never filled, and any item that needs a fifth state is recorded as a finding. +breaks: | + DDD-address-09, which defines a pull over these states, the examples of gaps it gives, and the state columns of DDD-address-06's index. +credits: | + The four state names, specified, actual, observed and expected, are GitHub user pjarnhus's, from his unpublished work on four states of an implementation (2026-11-18 R-6). He agreed to their appearing here under this credit (G1-Q7). The mapping of each state to canon is ours, made from the names. +owner: none +changed: 2026-11-18, act map, scope and pull session (G2, on the G1 rulings; DDD-dec-122) +notes: | + New on 2026-11-18. Ruled by Emil: 2026-11-18 R-5 and R-6 (meta/sessions/2026-11-18-pull-and- + scope/rulings.md) and G1-Q5 and G1-Q7 (meta/sessions/2026-11-18-act-map-scope-and-pull/ + rulings.md). + + Emil (2026-11-18 R-5): "the states are all that i have, but i completely agree with your + mapping to the canon, this is exactly what we need in order create a process for our + practicationers". + + G1-Q7: Emil told pjarnhus, and he agreed. This closes the open point of issue #80 that said + his agreement was not recorded. + + Upstream terms the mapping uses, settled at v5.15.2: term:governing-decision, term:outcome + ("the determinate as it lands in the world"), term:verdict ("the same determinate as assessed + by a declared predicate", in term:outcome) and term:tolerance. + + UNVERIFIED — Emil review, and open: + - The mapping is built from four words. pjarnhus's own definitions may differ. + - "Implementation" is a software word. The four states have not been tested as states of a + determinable outside software. diff --git a/core/decisions/DDD-dec-122.yaml b/core/decisions/DDD-dec-122.yaml new file mode 100644 index 0000000..cacd301 --- /dev/null +++ b/core/decisions/DDD-dec-122.yaml @@ -0,0 +1,139 @@ +format: 1 +id: DDD-dec-122 +kind: decision +statement: | + The repository defines the act map, its scope and its index (DDD-address-06), when a second act map exists (DDD-address-07), a process and north (DDD-address-08), the pull (DDD-address-09), kinds of demand as ground positions (DDD-address-10) and the four states (DDD-address-11), and method step 1 starts from scope and pull: the five-part pull procedure becomes the first part of step 1's card, [PROPOSED] inside a card ruled for test. Steps 2 to 10 each record what they add to the act map, [PROPOSED]. This adds to DDD-dec-114, amends DDD-dec-118 and supersedes the wording of two rulings of 2026-11-17, R-2 and R-7 point 3. +principal: Emil +made: | + 2026-11-17 and 2026-11-18, in the claude.ai project conversation (verbatim in meta/sessions/2026-11-17-act-map-and-process/rulings.md and meta/sessions/2026-11-18-pull-and-scope/prompt.md and rulings.md), and at G1 of issue #80 on 2026-11-18 at 10:30 and 13:26 Europe/Berlin (verbatim in meta/sessions/2026-11-18-act-map-scope-and-pull/rulings.md); filed by the act map, scope and pull session the same day; Emil's landing of the branch is the ratification act +resolution: | + ## Ruled on 2026-11-17 (2026-11-17 rulings.md) + + - **E-2. The layering is canon.** Acts at ground axes, with composite acts above and below, + and nothing else. + - **R-2. North.** North is a determinable, the cause, where demand originates. The discharge + act is the other end. Its wording, "the unresolved determinable", is superseded by K-1 below. + - **R-3. The index** is of determinables, with decisions as the ones governance took up. + - **R-4, R-5. A process** is a human understanding of work, not an object of the model. Its + translation is a named view, fixed by one or more norths and optionally bounded by composite + acts. In tooling a saved view, never a node. + - **R-6. Time** sets a determinable's rate of change, which is what creates demand. + - **R-7.** The map's nodes are act sites, with occurrences beside it. The index carries a rate + of change. A view leaves out earlier readings. Point 3's wording is superseded by G1-Q2. + Point 7, "no issue number on the commits", is superseded by issue #80 (2026-11-18 rulings.md, + first correction). + + ## Ruled on 2026-11-18, before the issue (2026-11-18 rulings.md) + + - **R-1.** A determinable has two or more outcomes that cannot all hold together. + - **R-2.** Scope is the first declaration of a map. Everything outside it is world, including + other people's decisions. The trace back from a request stops where the answer lies outside + the scope. + - **R-3.** A pull of a narrow scope appears in a wider map as a dependency, and as a seam only + where the wider map declares that scope. + - **R-4.** A kind of demand is a position on a declared ground axis. + - **R-5, R-6.** A pull is a gap between two of four states: specified, actual, observed, + expected. Credited to GitHub user pjarnhus. The mapping to canon is ours. + - **R-8.** The ids: DDD-address-06 to -08 and this decision keep their ids and are amended + before landing. New claims take DDD-address-09 onwards. + + ## Ruled at G1, 2026-11-18 (this session's rulings.md) + + - **C-1.** DDD-address-07's falsifier limb on applicability predicates holds only between + different declarations of axes. Nested scopes have their own limb. + - **C-2 and K-1.** A north is the determinable of a pull whether or not a gap is open now. This + supersedes the wording of 2026-11-17 R-2 and of the matching line in #80's "ruled before" + list, not their meaning. It rests on 2026-11-17 R-6. + - **Q-1.** DDD-address-06 as amended. **Q-2.** A second act map exists when a second scope or a + second declaration of ground axes exists. "Scope says what is world. Axes say what is + comparable." **Q-3.** North is the determinable of a pull. + - **Q-4.** Each map declares the axis its kinds are read on. The method offers "built or not + built" as the starting point. "Whose determinable" stays [PROPOSED]. + - **Q-5.** DDD-address-09 to -11 as written, conceptual and projected. + - **Q-6.** A pull is a gap between two of the four states. The gap table is examples, with + "expected against observed" added. "Specified against observed" is the readable stand-in for + "specified against actual". + - **Q-7.** DDD-address-11 credits pjarnhus by name. He agreed. + - **Q-8.** S-1: the act map is central as an index every later artefact attaches to, not as a + container. S-2, option A: the five parts become the first part of step 1's card. + - **Q-9 and K-2.** The nine "Value made visible" rows of steps 2 to 10, [PROPOSED]. Row 5 + attaches closure to the act site's acceptance predicate, closed for the arrangement allocated + there (term:closure). + - **Q-10.** The effects on T2b and on step 1b's kinds as stated in the session's effects.md. + DDD-address-02: a note appended, the region not reworded. + - **Q-11.** Filing as listed. + - **The reading of "found", and K-3.** A deviation is found once any holder observes an + outcome outside their tolerance: "expected against observed", found from the expected side. + Exposure nobody has observed is latent, which canon calls escaped (term:escape), and the + pull procedure does not find it. Finding it before it lands is assurance's work, in step 5. + + ## What is filed + + - DDD-address-06, -07, -08 (amended) and -09, -10, -11 (new), all projected. + - A note appended to DDD-address-02. + - Superseding sections, with nothing reworded, on the cards of steps 1, 1b and 2 to 10. + - This session's row in meta/sessions/README.md. + + ## What is not changed + + - No act, no source of pull and no other lifecycle data under applications/. + - Nothing in core/graph/, graph/upstream.yaml, the lc: vocabulary or the principle repository. + - The registers of the step cards: step 1 stays ruled for test, with a [PROPOSED] first part, + and steps 1b and 2 to 10 stay [PROPOSED]. + - T2b's pre-registration of 2026-09-27, which is fixed. Six items are registered again for the + T2b pre-registration not yet sealed (the session's effects.md, T-1 to T-6). + + ## Relation to earlier decisions + + - **Adds to DDD-dec-114.** Round 9's "the SDLC is a subprocess under the PDLC" keeps its ruled + reading, a composite act within the PDLC. The SDLC is a bound (DDD-address-08), and may be + declared a scope (DDD-address-06). + - **Amends DDD-dec-118.** Step 1b stays after step 1. Its option text "Acts first" is amended: + step 1 now starts from scope and pull. "Each process declares its own kinds" becomes "each map + declares its kinds, as positions on a declared ground axis" (DDD-address-08, -10). Its other + rulings stand. + - **Supersedes** no decision by the supersedes field, because neither earlier decision is + wholly replaced. It supersedes the wording of 2026-11-17 R-2 and R-7 point 3. + - **Release.** Not cut by this decision or this session. +basedOn: + - DDD-address-06 + - DDD-address-07 + - DDD-address-08 + - DDD-address-09 + - DDD-address-10 + - DDD-address-11 + - DDD-address-01 + - DDD-address-02 + - DDD-address-04 + - DDD-flow-01 + - DDD-flow-03 + - DDD-ground-05 + - DDD-ground-06 + - DDD-ground-01 + - DDD-frame-13 + - DDD-dec-114 + - DDD-dec-118 + - DDD-dec-38 +notes: | + History. First drafted on 2026-11-17 on a branch that did not land + (claude/act-map-and-process-2026-11-17; its commits kept on claude/pull-and-scope-2026-11-18 at + 47cf4cc), and amended on the rulings of 2026-11-18 (R-8) and of G1. The 2026-11-17 draft filed + three claims and two method sentences. Neither sentence is filed here: the card changes replace + them. + + Id. Checked on every branch of both repositories at arrival (08:51 UTC) and again before filing + (12:26 UTC), 2026-11-18. DDD-dec-122 and DDD-address-06 to -08 appear only on the archive branch + and the old branch. Nothing at DDD-address-09 or above exists anywhere. Upstream's highest + decision is DDD-dec-39. + + Corrections, with attribution, in the session's rulings.md: Emil's C-1 and C-2 on Claude's + proposals; Claude's catch of Emil's Q-9 reason against term:closure (K-2) and of his sentence + for step 1's card against term:escape (K-3). + + Open, not ruled here: step 5 classifies closure before step 6 allocates the arrangement it is + closed for (K-2). O-2 to O-5 and O-9 of 2026-11-17 stay open. Building the lifecycle's index of + determinables, and which step builds it, are out of #80's scope. + + Review trigger. A second mapped case, outside software: its act map either fits DDD-address-06, + its pulls the four states and its process the view of DDD-address-08, or not, and the six + claims are reviewed with it. diff --git a/meta/sessions/2026-11-17-act-map-and-process/bootstrap.md b/meta/sessions/2026-11-17-act-map-and-process/bootstrap.md new file mode 100644 index 0000000..3a6e349 --- /dev/null +++ b/meta/sessions/2026-11-17-act-map-and-process/bootstrap.md @@ -0,0 +1,54 @@ +# Bootstrap — the act map and a process, 2026-11-17 + +| Parameter | Value | +|---|---| +| Invocation | Interactive. Emil gave the issue text as one message in the claude.ai project conversation. `prompt.md` carries it verbatim | +| Principal | Emil (sole ratifier; ratifies by landing the branch) | +| Issue | None found on GitHub at arrival. The newest issue is #77, and none carries this title. The issue text exists only in the conversation and in `prompt.md`. Commits carry no `Refs` line until Emil names a number | +| Upstream repo | github.com/mindovermachine-dev/actor-indexed-determination, read only, at tag `v5.15.2` (`084bce7`) | +| Upstream pin | `v5.15.2` (`graph/upstream.yaml`, line 133), unchanged by this session | +| Downstream repo | github.com/mindovermachine-dev/decision-driven-design | +| Downstream base | `84aa466`, head of `main` at arrival | +| Downstream branch | `claude/act-map-and-process-2026-11-17` | +| Parallel work | PR #73 (`claude/zealous-ptolemy-3txf8k`, `applications/` for readers, `DDD-dec-94`, `-95`, `-119`) and PR #74 (`claude/confident-goldberg-c0y0ia`, the method plugin, `DDD-dec-121`). Both are open and unlanded. #73 rewrites prose under `applications/`, where this session's survey reads line numbers | +| Model | Claude Fable 5.2 | + +## Gates + +| Gate | What | Who | +|---|---|---| +| G0 | This charter committed before any other change | Claude | +| G1 | Deliverables 1 and 2: `survey.md` and `proposals.md`. Nothing is filed | Claude; **stops for Emil**, who rules each definition and where it is filed | +| G2 | Deliverable 3: the definitions filed as ruled; `python3 scripts/ci.py` passes; pushed | Claude; **stops for Emil** | +| Merge | Land the branch | **Emil** | + +## Scope at arrival + +In scope: the survey, the proposals, and after G1 the filings Emil rules. + +Out of scope, from the issue: changing any act; new processes such as budget or employment; Mantle; canon in the principle repository. + +## Decision ids + +Checked across every branch of both repositories at arrival (2026-11-17, 15:40 UTC). + +- **Upstream** (27 remote branches): the highest decision id is `DDD-dec-39`. +- **Downstream** (every `origin/*` branch): the highest decision file on `main` is `DDD-dec-118`. `DDD-dec-119` is on PR #73's branch. `DDD-dec-120` and `DDD-dec-121` are on PR #74's branch. No branch mentions an id above `DDD-dec-121`. + +The next free decision id is **`DDD-dec-122`**. Nothing is filed before G1. + +The highest claim id in the address area is `DDD-address-05`. + +## Deviations at arrival + +- **Reading before the charter.** The ground files were read, and the search for "act map" was run, before this charter was committed. Nothing was written. Recorded here because the convention puts the charter first. +- **Project instructions are stale on the head.** They name `main` at `4004b49`. `main` is at `84aa466`. The upstream pin they name, `v5.15.2`, is still the pin. + +## Read at arrival + +- `AGENTS.md`; `meta/way-of-working.md`; `meta/sessions/README.md`. +- `method/01-map-and-index-acts.md`; `method/quick-start.md` §3 and §4; `method/01b-map-the-flow-of-demand.md` (the lines that name the act map or a process). +- `core/claims/DDD-address-01.yaml`, `-02`, `-04`; `core/decisions/DDD-dec-114.yaml`, `-118`. +- `applications/pdlc/pdlc.md` §11 and §13; `applications/sdlc/sdlc.md` §3 and §10; `applications/pdlc/graph/ontology.ttl`, `shapes.ttl`, `acts.ttl`; `applications/sdlc/acts/acts.ttl`. +- `meta/sessions/2026-09-27-act-address/` (`rulings.md`, `prereg-act-coordinates.md`, `results-act-address-dry-run.md`). +- Upstream at `v5.15.2`: `core/claims/DDD-ground-06.yaml`; `term:act` and `term:act-address` in `core/graph/terms.yaml`; `core/14-indexed-determination.md` §5. diff --git a/meta/sessions/2026-11-17-act-map-and-process/predictions.md b/meta/sessions/2026-11-17-act-map-and-process/predictions.md new file mode 100644 index 0000000..76dc2cc --- /dev/null +++ b/meta/sessions/2026-11-17-act-map-and-process/predictions.md @@ -0,0 +1,28 @@ +# Predictions — the act map and a process, 2026-11-17 + +Written before each operation is run. Results are recorded against each prediction. + +## P-1 — the survey + +Not a prediction. The search for "act map" ran before the charter was committed (`bootstrap.md`, deviations). Its result could not be predicted afterwards, so none is claimed. + +**Result:** 109 lines in 38 files. See `survey.md`. + +## P-2 — the full gate at G1 + +`python3 scripts/ci.py` passes on this branch at G1. The branch adds session records only. It changes no claim, no decision, no What or How file and no fixture. Two environment failures are likely before any gate reads a file, as in `2026-11-08-specification-feasibility/predictions.md` P-2: `ssh-keygen` missing, and the shallow clone hiding release tags. + +**Result:** held, with one of the two environment failures. The first run failed at `preflight`: `ssh-keygen` was not on PATH. Installed (`openssh-client`). Tags were fetched before the run (`git fetch --force --tags origin`), so the releases gate was not seen to fail. The second run passed every gate (exit 0). + +## P-3 — the full gate at G2 + +Written at 21:27 (Europe/Berlin), after R-7 and before any filing. + +Filed: `core/claims/DDD-address-06.yaml`, `-07`, `-08`; `core/decisions/DDD-dec-122.yaml`; one sentence in `method/01-map-and-index-acts.md` and one in `method/01b-map-the-flow-of-demand.md`. + +1. `python3 scripts/validate-claims.py core/claims/` passes. It may warn on rule 1 for `DDD-address-06` and `-08`, whose statements list parts. A warning is not an error (`spec/claim-format.md` §4). +2. `python3 scripts/validate-claims.py core/decisions/ --decisions` passes. Every id in `DDD-dec-122`'s `basedOn` exists here or is pinned in `graph/upstream.yaml` (`DDD-ground-06` and `DDD-frame-13` are). +3. `python3 scripts/ci.py` passes. No pin moves, no release descriptor is added, and no What, How or lifecycle file is touched. +4. The ids are free. Re-checked at 21:26 on every branch of both repositories: `DDD-dec-122` and `DDD-address-06` to `-08` appear only in this session's own records. + +**Result:** held, with one miss on the cautious side. (1) Passed, and no rule 1 warning fired on the three new claims: the six warnings are on claims this session did not touch. (2) Passed: 97 decisions. (3) Passed: every gate, exit 0, after the session index row was added. (4) Held. diff --git a/meta/sessions/2026-11-17-act-map-and-process/prompt.md b/meta/sessions/2026-11-17-act-map-and-process/prompt.md new file mode 100644 index 0000000..6078a84 --- /dev/null +++ b/meta/sessions/2026-11-17-act-map-and-process/prompt.md @@ -0,0 +1,21 @@ +# The repository says what an act map is, when there is more than one, and what a process is + +The issue, verbatim, as Emil gave it in the claude.ai project conversation on 2026-11-17 at 16:27 (Europe/Berlin). + +> The repository says what an act map is, when there is more than one, and what a process is +> +> Outcome. One filed definition each of the act map and of a process, and a stated rule for when a second act map exists. +> +> Ground. `AGENTS.md`. `meta/way-of-working.md`. `method/01-map-and-index-acts.md`. `core/claims/DDD-address-01.yaml` and `DDD-address-02.yaml`. `applications/pdlc/pdlc.md` §11. `applications/sdlc/sdlc.md`. `applications/pdlc/graph/ontology.ttl`. Canon at the tag in `graph/upstream.yaml`. +> +> Ruled before this issue was opened, 2026-11-17, in Mantle’s planning project: for Mantle’s view, a process is a composite act that holds acts directly. Whether this repository adopts that is ruled here. +> +> Deliverable. +> +> 1. Every use of “act map” in the repository, with what each implies: one or many, tree or graph. Exit: the list names each file and line. +> 2. Proposed definitions, each with a falsifier and what it breaks: the act map; when a second one exists; a process. Exit: each says which filed claim or ruling it rests on, or that it rests on none. +> 3. On the rulings, the definitions are filed where the session proposed and the principal ruled. Exit: `python3 scripts/ci.py` passes. +> +> Gates. G1 after step 2: the principal rules each definition and where it is filed. G2 after step 3. Then the merge. +> +> Out of scope. Changing any act. New processes such as budget or employment. Mantle. Canon in the principle repository. diff --git a/meta/sessions/2026-11-17-act-map-and-process/proposals-2.md b/meta/sessions/2026-11-17-act-map-and-process/proposals-2.md new file mode 100644 index 0000000..303729f --- /dev/null +++ b/meta/sessions/2026-11-17-act-map-and-process/proposals-2.md @@ -0,0 +1,161 @@ +# Proposals, second form — after R-2, R-3 and R-4 + +The three definitions as they stand after Emil's rulings of 2026-11-17. This file supersedes `proposals.md`, which is kept as written. Nothing here is filed. + +| # | Definition | Ruled so far | Still to rule | +|---|---|---|---| +| D-1 | **The act map** is the record of one ruled decomposition under one declaration of ground axes. It holds act sites, the composite acts above and below them, the dependencies between them, and the index of determinables | Layers are composite acts and nothing else (E-2). The index is of determinables (R-3) | Nodes are sites. "Top act". Whether the index is part of the map | +| D-2 | **A second act map** exists when, and only when, a second declaration of ground axes exists | Nothing | The whole rule | +| D-3 | **A process** is a human understanding of work, scoped to an area of expertise. Its translation into the model is a view of an act map: the acts between one unresolved determinable and the determinate that discharge produces | North is the unresolved determinable; the discharge act is the other end (R-2). A process is a human concept, and this is its best translation (R-4) | The wording. Whether the view leaves out earlier readings | + +Arguments below that Emil has not confirmed are `UNVERIFIED — Emil review`. + +## D-1 — the act map + +**Definition.** The act map is the record of one ruled decomposition of work into acts, read under one declaration of ground axes. It holds four things: + +| Part | What it is | Source | +|---|---|---| +| Act sites | One site per address: a ground position on the declared axes and a layer path. The occupant is not part of it | `DDD-ground-06` | +| Layers | Composite acts above an act and below it. There is no other kind of layer. A top act is a composite act with none above it | `DDD-ground-06`; `core/09` §1; E-2 | +| Dependencies | Site B depends on site A when a state change an occurrence at A produced is in the ground reading of an occurrence at B. A directed graph | `DDD-address-01` | +| The index of determinables | Every determinable that makes demand arrive, with the act where it enters and the act where its determinate lands. Each is marked taken up, which makes it a decision anchored to acts, or not taken up, which makes it a candidate escape | `DDD-frame-13`; `core/00` §4a; R-3 | + +The map has no direction of its own beyond causal order. A view gives it one by naming a north: one determinable from the index (E-1, R-2). + +Occurrences are not in the map. Each is recorded against one site, with its time index. + +**Weakest point.** Two, of different kinds. +1. `DDD-address-02`'s region says "a tree of occurrences". This says the nodes are sites. Unchanged from `proposals.md`. +2. The index is new, and nothing in the repository has one. The lifecycle's six sources of pull are the nearest thing: each names an entry act and a discharge act, and none is marked taken up or not. So the fourth part is defined before it has been built once. + +**Rests on.** `DDD-ground-06`, `DDD-address-01`, `DDD-address-02`, `DDD-frame-13` (all projected); `core/00` §4a; the `lc:` vocabulary (Round 28, R30); E-2 and R-3 of this session. "Top act" and "occurrences sit outside the map" rest on **no claim**. + +**Falsifier.** Any of: +1. An act site that must sit within two composite acts for its governing set to be stated. +2. A dependency that can be stated between two occurrences and between no two sites. +3. A determinable that makes demand arrive and that cannot be given an entry act on the map. This limb also yields a finding: work the decomposition does not cover. + +**What it breaks.** Limb 1: the layer path and `DDD-ground-06`'s "exactly one address". Limb 2: the split between map and occurrence record. Limb 3: the claim that the map shows the pull at all. + +## D-2 — when a second act map exists + +Unchanged from `proposals.md`, and not yet ruled. + +**Rule.** A second act map exists when, and only when, a second declaration of ground axes exists. A second top act under the same declaration is a second tree in the same map. A view is not a second map. + +**Weakest point.** Whether two pieces of work share a declaration is the mapper's decision. The rule names it and does not make it. + +**Rests on.** `DDD-ground-06`'s region; `DDD-address-04`; the accepted T2a map (`DDD-dec-38`), which has three roots under one declaration. That the declaration is the boundary rests on **no filed claim or ruling**. + +**Falsifier.** A governing decision whose applicability predicate must cover occurrences in two maps, or two occurrences in two maps whose behaviour must be compared. + +**What it breaks.** Checking a map on its own, and the two grains of `DDD-address-05`. + +## D-3 — a process + +**Definition.** A process is a human understanding of work being done. It is scoped and settled within an area of expertise, such as software, finance or HR. It is not an object of the model. + +Its translation into the model is a view of an act map, fixed by one north: + +| Part of the view | What it is | Source | +|---|---|---| +| North | One unresolved determinable: the cause, where demand originates. It exists whether or not anyone has taken it up | `DDD-frame-13`; `core/00` §4a; R-2 | +| The other end | The act at which discharge produces the determinate | `core/09` §7a; R-2 | +| The acts shown | The act sites on a dependency path from the determinable's entry act to that discharge act | `DDD-address-01` | +| Rank | Distance from north along dependencies | E-1 | + +The translation is the passage from determinable to determinate (`core/00` §4a), read over a chain of acts. It may cross composite acts. + +**Weakest point.** A process is settled inside one area of expertise, so there is no fact against which the translation can be wrong. It can only be more or less useful to the people who hold the process. The falsifier below is therefore a test of utility, as `DDD-address-02`'s is, and weaker than a test of fact. + +**What this settles without a new ruling.** +- **Round 9.** "The SDLC is a subprocess under the PDLC" is software's own word for its work. Its ruled reading in the model, a composite act within the PDLC, stands (`DDD-dec-114`). Nothing is superseded. +- **The lifecycle.** The six sources of pull are six norths. The SDLC is a composite act that several views pass through. +- **Mantle.** "A composite act that holds acts directly" is another translation of the same human concept, made for display. Whether Mantle keeps it is ruled in Mantle's project. + +**Rests on.** R-2 and R-4 of this session; `DDD-frame-13`; `core/00` §4a; `core/09` §7a; `DDD-address-01`. "Process" has no filed term, here or upstream. + +**Falsifier.** Either: +1. A process its holders recognise that has no north: no determinable in the world starts it. Work that starts only because an earlier decision scheduled it is the candidate. The holding note `meta/holding-note-addendum-determinables.md` Q36 (upstream, not ratified) calls that ledger-pull. +2. A view that the holders of the process do not recognise as their process, in a second mapped case. + +**What it breaks.** Step 1b's use of "a process" as its unit, and the promotion condition of `DDD-flow-01` to `-04` ("a second process"). + +## Where to file — proposed, not ruled + +| What | Where | Form | +|---|---|---| +| The rulings of this session | `core/decisions/DDD-dec-122.yaml` | One decision. `basedOn` the claims below, `DDD-ground-06`, `DDD-frame-13`, `DDD-address-01`, `DDD-dec-114`, `-38` | +| D-1, the act map | `core/claims/DDD-address-06.yaml` | Conceptual, projected. Test: T2b | +| D-2, a second act map | `core/claims/DDD-address-07.yaml` | Normative, projected | +| D-3, a process | `core/claims/DDD-address-08.yaml` | Normative, projected | +| Pointers | `method/01-map-and-index-acts.md`, "Artefact out"; `method/01b-map-the-flow-of-demand.md`, first use of "process" | One cited sentence each | + +Not proposed: the term registry, the `lc:` vocabulary, upstream, and any change to the lifecycle's acts or sources of pull. + +## What the rulings would leave open + +| # | Item | +|---|---| +| O-1 | `DDD-address-02`'s region says "a tree of occurrences" | +| O-2 | `sdlc.md` §10: the second top act has no graph data, and one sentence contradicts `DDD-ground-06` (S-8, S-9) | +| O-3 | Four places call the SDLC's part "the act map" (S-10) | +| O-4 | The o5 and o6 loop is left out of the data (S-4) | +| O-5 | New demand declares oC as an entry act, and no dependency leads from oC towards o16 for new demand (`rulings.md`, evidence) | +| O-6 | The lifecycle has no index of determinables. Its six sources of pull are not marked taken up or not. Building the index changes the lifecycle data, which this issue puts out of scope | +| O-7 | Which step builds the index: step 1, or step 3, which is not filed | +| O-8 | PR #73 rewrites prose under `applications/`. The survey's line numbers are `main`'s at `84aa466` | +| O-9 | Carry R-2 and R-4 to Mantle's planning project | + +## For G1 + +1. D-1: are the map's nodes act sites, with occurrences recorded beside it? +2. D-1: is the index of determinables a part of the act map? +3. D-2: is the rule ruled as written? +4. D-3: is the wording ruled as written? Does a view leave out earlier readings (`lc:readsEarlier`), so that rank is strict? +5. Filing: one decision and three claims in the address area? +6. Scope: definitions only in this session, with O-6 and O-7 as new issues? +7. Is there a GitHub issue number for the commits? + +## Amendments + +Added after the file was first committed. The sections above are left as written. + +### A-1 — D-3 after R-5 and E-5 + +**Definition, as amended.** A process is a human understanding of work being done, scoped and settled within an area of expertise. It is not an object of the model. Its translation into the model is a **named view** of an act map, fixed by: + +| Part | What it is | How many | +|---|---|---| +| Norths | Unresolved determinables: the causes the work answers | One or more | +| Bound | Composite acts the view is limited to | None, one or more | + +The view shows the act sites on a dependency path from each north's entry act to the act where its determinate lands, inside the bound where one is given. + +**Why two parts.** People name a process by where the work happens and who holds it, and by why it happens. The model keeps those apart: the first is composite acts, the second is norths. The SDLC is a bound with the norths that end at its discharge act. The PDLC is one bound with six norths. + +**In tooling.** A process is a saved view with a name. It is never a node, so no decision can be anchored to it. + +**Falsifier, as amended.** Limb 1 of D-3 ("a process with no north", with calendar-started work as the candidate) is withdrawn on E-5: time makes a determinable unresolved again, so calendar-started work has a north. What remains: + +1. A view that the holders of the process do not recognise as their process, in a second mapped case. +2. A process whose holders cannot name any determinable it answers, after the calendar is set aside. + +**Weakest point, added.** If time is the cause of every determination, then "time" cannot tell one north from another. What tells them apart is which determinable time moves, and how fast. So north stays the determinable, and a north's rate of change is a property the index may need. Nothing filed says so. + +**Rests on.** R-2, R-4, R-5 and E-5 of this session; `DDD-frame-13`; `core/00` §4a; `DDD-address-01`; `DDD-ground-06` for the bound. + +**Effect on Mantle.** "A composite act that holds acts directly" is a view with a bound and no north. It is a case of this definition, not a rival to it. Whether Mantle keeps it is ruled there (O-9). + +### A-2 — D-1 after R-6 + +**The index of determinables, as amended.** Each determinable in the index carries a fourth property beside its entry act, its discharge act and whether it is taken up: its **rate of change**, the rate at which time makes it unresolved again. That rate is what creates demand (R-6). + +**What it gives.** A north's rate of change says how often the work recurs. The lifecycle writes this on acts, as free text (`lc:recurs "per issue"`, `"per event"`). On R-6 it is a property of the cause first, and the acts inherit it. + +**Weakest point.** No filed claim carries a rate of change for a determinable. The method's step 4, which is `[PROPOSED]` and not filed, speaks of declaring ground with its tick rate. So this property is defined here before it has a unit or a way to measure it. + +**Falsifier.** A determinable whose demand recurs at a rate that its own rate of change does not explain, for example work that recurs because an act upstream failed and not because the world moved. That is owed demand in the lifecycle, and it may need a rate of its own. + +**Rests on.** R-6 and E-5 of this session; `DDD-frame-13`. On no filed claim for the rate itself. diff --git a/meta/sessions/2026-11-17-act-map-and-process/proposals.md b/meta/sessions/2026-11-17-act-map-and-process/proposals.md new file mode 100644 index 0000000..09e07cf --- /dev/null +++ b/meta/sessions/2026-11-17-act-map-and-process/proposals.md @@ -0,0 +1,139 @@ +# Proposals — the act map, a second act map, a process + +Three definitions, `[PROPOSED]`. Nothing here is filed. Emil rules each one, and where it is filed, at G1. + +| # | Proposal | Rests on | Weakest point | +|---|---|---|---| +| D-1 | **The act map** is the record of one ruled decomposition under one declaration of ground axes. It holds act sites. It has two relations: containment, which is a tree under each top act, and causal dependency, which is a directed graph | `DDD-ground-06`; `DDD-address-01`; `DDD-address-02`; the `lc:` vocabulary (Round 28, R30) | It says "sites" where `DDD-address-02`'s region says "occurrences" | +| D-2 | **A second act map** exists when, and only when, a second declaration of ground axes exists. A second top act under the same declaration is a second tree in the same map | `DDD-ground-06`'s region; `DDD-address-04`; the accepted T2a map | Whether two pieces of work share a declaration is the mapper's decision. The rule names the decision. It does not make it | +| D-3 | **A process** is a composite act. Mantle's ruling is its limiting case, with depth fixed at one | `DDD-dec-114` (Round 9, ruled reading); `DDD-ground-06` | It adds a second word for a filed term and no new content | + +Every argument below that Emil has not confirmed is `UNVERIFIED — Emil review`. That is all of them. + +## D-1 — the act map + +**Definition.** The act map is the record of one ruled decomposition of work into acts, read under one declaration of ground axes. It holds: + +| Part | What it is | Shape | Source | +|---|---|---|---| +| Act sites | One site per full address: a ground position on the declared axes and a layer path. The occupant is not part of it | Nodes | `DDD-ground-06` | +| Containment | Each act site and each composite act sits within at most one composite act. A composite act within none is a **top act** | A tree under each top act. A forest when there is more than one top act | `DDD-ground-06` (the layer path); `DDD-address-02` (no act is shared) | +| Dependency | Site B depends on site A when a state change an occurrence at A produced is in the ground reading of an occurrence at B | A directed graph. Not a tree | `DDD-address-01` | + +Occurrences are not in the map. Each occurrence is recorded against exactly one site, with its time index. The record of occurrences is a second artefact. + +**Weakest point.** `DDD-address-02`'s region and step 1's card say "a tree of occurrences". This definition says the nodes are sites. The claim's statement is untouched: every occurrence is still its own act, and no act has two parents. But a ruled region is being read more narrowly than it is written, and ruled material changes by supersession. If Emil holds that the map's nodes are occurrences, then the lifecycle's `acts.ttl` is not an act map, because `lc:Act` is a site (S-5). + +**Why "tree or graph" has two answers.** The survey finds 11 lines that need a tree and 13 that need a graph. They are about different relations. "A tree of occurrences" (R-6) rules out one act under two parents. "Follows an edge of the act map" (`lc:` shapes) walks dependencies. Neither can be given up: the layer path needs the tree, and step 1b and the fan-out query need the graph. + +**Rests on.** +- `DDD-ground-06` (upstream, projected): the address, the layer path, and "addresses are relative to a ruled decomposition". +- `DDD-address-01` (projected): dependency over ground readings. +- `DDD-address-02` (projected): every occurrence is its own act. +- The `lc:` vocabulary, ruled by Round 28 (R30): `lc:Act` as a site, `lc:within`, `lc:dependency`. +- "Top act" rests on **none**. It is used in `sdlc.md` §3 and §10 and is not filed (S-8). This definition would file it. +- That occurrences sit outside the map rests on the `lc:` data and on **no claim**. + +**Falsifier.** Any of: +1. An act site that must sit within two composite acts for its governing set to be stated. +2. A dependency that can be stated between two occurrences and between no two sites. +3. A question about the work that the sites and the occurrence record together cannot answer, and a single tree of occurrences can. + +**Pressure case already in the repository.** S-4: a later o5 reads an earlier o6, and o6 reads o5. Between sites that is a loop. The data omits one edge to keep sites acyclic. Limb 2 does not fire, because the dependency can be stated between sites. But "dependencies are acyclic" is true of occurrences and not of sites, and the definition must not claim otherwise. + +**What it breaks.** If limb 1 fires: `lc:ActShape`, the uniqueness of the layer path, and `DDD-ground-06`'s "exactly one address". If limb 2 or 3 fires: the split between map and record, and with it the lifecycle graph data and Mantle's reading of it. + +## D-2 — when a second act map exists + +**Rule.** A second act map exists when, and only when, a second declaration of ground axes exists. Two act sites belong to one map exactly when their ground positions are read on the same declared axes, under the same ruled decomposition. + +| Case | One map or two | Why | +|---|---|---| +| The PDLC's acts and the SDLC's acts, in two files | One | One declaration: surface read, change locus, authority read (`DDD-dec-115`) | +| A second top act under the same axes, such as changes to the SDLC itself (`sdlc.md` §10) | One, with two trees | Same declaration. T2a already has three roots in one accepted map (S-7) | +| The quick start's Varve slice and the lifecycle | Two | Different declarations | +| The same work, decomposed a second way | One at a time | Re-decomposition supersedes (`DDD-ground-06`'s region). The old map is history | +| A part of a map, such as o1b to o13 | Not a map | It is the acts within one composite act of the map (S-10) | + +**Why the axes decide.** An address is a position on declared axes. Coincidence is equality of ground positions (`DDD-address-04`). Applicability predicates range over positions (`DDD-ground-01`). None of the three is defined between two declarations. So the boundary of a map is the boundary of what can be compared and governed together. Between two maps only one relation remains: an act in one reads state that an act in the other produced. + +**Weakest point.** Whether two pieces of work share a declaration is decided by whoever rules the decomposition. Any two maps can be made one by declaring the union of their axes. Then some acts never read some axes, which is the held T2b test item F-1: "an act that never reads an axis a governing predicate uses" (act address landing, GATE 1 ruling 7). Whether that is a position is not ruled. So the rule says what a second map is and what it costs. It does not say when to make one. + +**The alternative, for the ruling.** One map per top act. It is simpler to state, and it matches "the SDLC's act map" in four places. It fails on the accepted T2a map, which has three roots (S-7). It would also make the changes to the SDLC a second map whose acts are compared with the first at every o11, across a boundary where comparison is not defined. + +**Rests on.** +- `DDD-ground-06`'s region: axes are declared, and addresses are relative to a ruled decomposition. +- `DDD-address-04`: coincidence at ground position. +- Act address landing, GATE 1 ruling 7 (`DDD-dec-38`): the T2a map accepted. +- `DDD-dec-115`: the lifecycle map's three axes. +- That the axis declaration is the boundary rests on **no filed claim or ruling**. It is this session's reading of S-6. + +**Falsifier.** Either: +1. A governing decision whose applicability predicate must cover occurrences in two maps. +2. Two occurrences in two maps whose behaviour must be compared (`DDD-address-05`) to answer a question the work poses. + +Each shows a split that loses governance. The second application case will test it: an organisation mapped beside the PDLC either shares the PDLC's axes or does not. + +**What it breaks.** Checking a map on its own (`check.py`). The two grains of `DDD-address-05`, which assume one address space. Any tool that points from one map at acts in another. + +## D-3 — a process + +**Definition.** A process is a composite act. A subprocess is a composite act within another. A process's acts are the act sites whose layer paths pass through it. + +**Mantle's ruling as the limiting case.** "A process is a composite act that holds acts directly" is this definition with depth fixed at one: the members shown are the acts one step below. Mantle's view may keep it. This repository cannot adopt it as written: + +| Composite act | Holds acts directly | A process under Mantle's ruling | Called a process in this repository | +|---|---|---|---| +| PDLC | Yes: o0, o2a, oS, o1, oC, oR, o16, oU | Yes | Yes (`DDD-dec-118`; step 1b) | +| SDLC | **No.** It holds frame, build, accept, land and run | **No** | **Yes.** Round 9: "the SDLC is a subprocess under the PDLC". Round 5: "the process of getting changes to production" (`DDD-dec-114`) | +| frame, build, accept, land and run | Yes | Yes | No. They are "stages" (`acts.ttl`) | + +Adopting the ruling as written would un-name the SDLC as a process, or require moving its acts, which is out of scope. + +**Weakest point.** The definition adds nothing canon lacks. "Process" becomes a reader's word for "composite act". If Emil means more by a process (owners, declared kinds of demand, an entry and a discharge), this definition is too weak, and step 1b is where the stronger one would come from. I found no ruling that asks for more. + +**The alternative, for the ruling.** A process is a composite act that declares kinds of demand, with an entry act and a discharge act for each (step 1b's unit). It separates the PDLC from "frame". It rests on `DDD-dec-118` and on four projected claims with one process as their whole evidence, so it would define a general word from one case. + +**Rests on.** +- `DDD-dec-114`: Round 9, with the reading ruled in Round 14, already equates a subprocess with a composite act within another. +- `DDD-ground-06` and `term:act-address`: composite acts and the layer path. +- "Process" itself rests on **no filed claim or term**, here or upstream. +- Mantle's ruling is taken from the issue text. This session has not read Mantle's planning project. + +**Falsifier.** A use of "process" in this repository, or in a mapped case, that names something no composite act can be: work with no containing act, or a flow that crosses two top acts and belongs to neither. + +**What it breaks.** Step 1b's wording ("each process declares its own kinds of demand") and the promotion condition of `DDD-flow-01` to `-04` ("a second process"), which would then have no defined subject. + +## Where to file — proposed + +| What | Where | Form | +|---|---|---| +| The rulings of G1 | `core/decisions/DDD-dec-122.yaml` | One decision, `basedOn` the claims below and `DDD-dec-114`, `-115`, `-38` | +| D-1, the act map, with "top act" | `core/claims/DDD-address-06.yaml` | Conceptual, projected. Test: T2b | +| D-2, a second act map | `core/claims/DDD-address-07.yaml` | Normative, projected, as `DDD-address-02` is: a governing choice about representation | +| D-3, a process | `core/claims/DDD-address-08.yaml` | Normative, projected | +| Pointers | `method/01-map-and-index-acts.md`, "Artefact out" row; `method/01b-map-the-flow-of-demand.md`, first use of "process" | One cited sentence each | + +Not proposed: +- **The term registry.** `core/graph/terms.yaml` holds terms established by numbered core documents. None of the three is. +- **The `lc:` vocabulary.** It is ruled, and an `lc:Process` class would add a second name for `lc:CompositeAct`. +- **Upstream.** Out of scope. The act map is diachronic, so it files downstream by `DDD-dec-09`'s charter in any case. + +## What the rulings would leave open + +| # | Item | From | +|---|---|---| +| O-1 | `DDD-address-02`'s region says "a tree of occurrences". If D-1 is ruled, the region is superseded or annotated, not rewritten | D-1 | +| O-2 | `sdlc.md` §10: the second top act has no graph data, and "the same addresses with a different layer path" contradicts `DDD-ground-06` | S-8, S-9 | +| O-3 | Four places call the SDLC's part "the act map" | S-10 | +| O-4 | The o5 and o6 loop is left out of the data | S-4 | +| O-5 | PR #73 rewrites prose under `applications/`. The survey's line numbers are `main`'s at `84aa466` | `bootstrap.md` | + +## For G1 + +1. D-1: are the map's nodes act sites, with occurrences in a record beside it? +2. D-1: is "top act" filed, as a composite act within no other? +3. D-2: does the axis declaration bound a map, or does each top act have its own? +4. D-3: is a process any composite act, with Mantle's ruling as the depth-one view? +5. Filing: one decision and three claims in the address area, or another form? +6. Is there a GitHub issue number for the commits? diff --git a/meta/sessions/2026-11-17-act-map-and-process/rulings.md b/meta/sessions/2026-11-17-act-map-and-process/rulings.md new file mode 100644 index 0000000..5427f5e --- /dev/null +++ b/meta/sessions/2026-11-17-act-map-and-process/rulings.md @@ -0,0 +1,39 @@ +# Rulings — the act map and a process, 2026-11-17 + +Emil's rulings and statements, in the order given, in his words. Recorded when given. G1 closed at R-7. `proposals-2.md`, with its amendments A-1 and A-2, carries the definitions as ruled. + +| # | Time (Europe/Berlin) | Status | Emil's words | Reading | +|---|---|---|---|---| +| E-1 | 19:28 | Direction given, not a ruling | "im thinking that an act map must have a direction - and ours must have acts and ground axes. Then we must take into account what a process is, and how that is a simple view of our act map. Because ours have layers and act chains. - a full act map will show all the acts at their ground axes, which will be a wierd visual picture. I think the direction needs to be the demand pull, so "north" is a demand pull, and then we can change the diretion to get different views. This will also allow us to see the chain of acts to a pull. Which then leads us to the conclussion that a process is an act map with a direction towards the defined demand pull" | A process is a view of the act map, directed at a declared demand pull. Claude's reading, `UNVERIFIED — Emil review` | +| E-2 | 19:31 | Statement of canon | "We have the layering its the acts at ground axes and then we have composite acts below and above that - it can never be anything else. We have ruled this in the actor indexed determation theory repo" | Checked at `v5.15.2`: `DDD-ground-06` (the layer path is "the chain of composite acts it sits within"); `core/14` §5 and `core/09` §1 ("acts nest as actors nest"). A layer is a composite act and nothing else | +| R-1 | 19:34 | **Ruled** | "yes to the first" | North is the discharge act of a declared demand kind | +| — | 19:34 | Open, asked by Emil | "but for the other end im not sure, could there be multiple ends ?" | Not ruled. See below | +| E-3 | 20:27 | Direction given. Whether it is a ruling is asked | "we follow the determination pull, which means that the decisions that must be made, this also means that we must index the decisions not just the acts. without that index how can we see the pull" | The pull is determination demand: the governing decisions that must be made. Checked at `v5.15.2`: `term:act` ("an act is decisions resolved against ground"; "the unit demand is counted in"), `term:conservation` ("Every governing decision gets made"), `term:last-decision`. The lifecycle map carries no decision on any act: an act has a label, an address, a holder and dependencies. So R-1's north, the discharge act of a demand kind, is a stand-in for a decision. Claude's reading, `UNVERIFIED — Emil review` | +| E-4 | 20:43 | Correction by Emil. Whether it supersedes R-1 is asked | "North cant be a decision it must be where the demand for decisions originate, so it is a cause. This is very clear in the actor theory, a fire in a forest creates demand even though no one is there. Take a look at the Canon we have defined this befire" | Checked at `v5.15.2`. `DDD-frame-13` (projected): "determination demand is thereby a measure over unresolved determinables". `core/00` §4a: "a decision is a determinable taken up by governance", and "what no governance takes up is still a determinable — resolved, when its act completes, by whatever resolves it". `core/09` §7a: a determinate lands "whether or not anything governed it". The fire itself is in `meta/holding-note-addendum-determinables.md` Q36 ("World-pull … the fire"), which is a holding note and not ratified. North is therefore an unresolved determinable: the cause. Claude's reading, `UNVERIFIED — Emil review` | +| R-2 | 21:12 | **Ruled**, supersedes R-1 | "1. Yes i think so this would make the most sense." | Asked: "Is north the unresolved determinable, the source of pull, and does that supersede R-1? R-1's discharge act would become the other end." North is the unresolved determinable. The discharge act is the other end | +| R-3 | 21:12 | **Ruled** | "2. Yes - that would build the index we need." | Asked: "Is the index an index of determinables, with decisions as the ones governance took up?" The index is of determinables | +| R-4 | 21:12 | **Ruled** | "3. A process is a simple human understanding og work being done, it is scoped and settled usually in a way that only makes sense within a given area of expertise. Software, finance, hr etc. What we have is a much more complex model of the world and mapping a process into our model is going to be hard. But I still think its the best translation of the concept we have" | A process is a human concept, scoped to an area of expertise. It is not an object of the model. The view from determinable to determinate is its translation into the model, and the best one available. Claude's reading, `UNVERIFIED — Emil review`: a process may therefore cross composite acts, and that is not a defect in the decomposition | +| R-5 | 21:19 | **Ruled** | "agreed on all" | Said in reply to Claude's view of 21:15, which proposed three things. (1) A process translates to a named view with one or more norths, optionally bounded by composite acts, because people name processes by scope and ownership as well as by cause. (2) In tooling a process stays a saved view with a name, never a node. (3) The second mapped case is where the translation is tested. Claude reads "all" as these three, and not as the seven G1 questions of 21:12. `UNVERIFIED — Emil review` | +| E-5 | 21:19 | Statement | "the calendar observation is good, time is always the cause of determination one way or the other the law of thermo dynamics is in essence what will ensure that the world changes with time." | Work that starts on a calendar has a north like any other: a determinable that time makes unresolved again. The schedule is governance sampling a world that changes whether or not anyone looks. Not checked against a filed claim: the nearest text is the upstream holding note's Q36 ("the world sets the tempo and never misses a tick"), which is not ratified. Claude's reading, `UNVERIFIED — Emil review` | +| R-6 | 21:22 | **Ruled** | "agreed time is the one that sets the change rate of a determinable which is what creates the demand" | Said in reply to Claude's caution of 21:19, that time cannot tell one north from another and that what tells them apart is which determinable time moves and how fast. Time sets a determinable's rate of change, and that change is what creates demand. North stays the determinable. Claude does not read this as closing G1: the seven questions were not answered. `UNVERIFIED — Emil review` | +| R-7 | 21:25 | **Ruled. G1 closed** | "i agree and accept on all seven points" | The seven points as Claude listed them at 21:19, with the rate of change added to point 2 at 21:22. (1) The map's nodes are act sites, with occurrences recorded beside it. (2) The index of determinables is a part of the act map, and each determinable carries its rate of change. (3) D-2 as written: a second act map exists only with a second declaration of ground axes. (4) D-3 as amended in A-1, and a view leaves out earlier readings, so rank from north is strict. (5) One decision, `DDD-dec-122`, and three claims, `DDD-address-06` to `-08`, with one cited sentence each in `method/01` and `method/01b`. (6) Definitions only in this session; building the lifecycle's index, and which step builds it, become new issues. (7) No issue number on the commits | + +## Corrections + +- **Emil, on Claude's reading of E-3 (E-4).** Claude proposed that north is "a decision that must be made". That fused demand with decisions. Canon keeps them apart: demand is over determinables, and a decision is a determinable that governance has taken up. Demand exists before any decision, and without any actor. The proposal is withdrawn. + +- **Claude, on E-2.** `proposals.md` D-3 proposed "a process is a composite act". After E-2 that proposal is withdrawn: the only name for a layer is composite act, so a process cannot be one. Claude's question whether to keep "two words, one for the layer and one for the chain" is withdrawn with it. `proposals.md` is revised at the close of G1, not rewritten now. + +## Evidence computed for the open question + +Run over `applications/pdlc/graph/acts.ttl`, `applications/sdlc/acts/acts.ttl` and `demand-flow.ttl` at `84aa466`. Dependencies only (`lc:dependency`), honouring `lc:forKind`. Earlier readings (`lc:readsEarlier`) are left out. + +| Demand kind | Declared entry acts | North | Acts behind north | Where the chain ends on its own | +|---|---|---|---|---| +| New | o0, oC | o16 | 15 | o0 | +| Owed | oC | o13 | 11 | oC | +| Own-ground | oG | o13 | 11 | oG | + +- For each kind, every act behind north is also after a declared entry act. Nothing is behind north and outside the entry. +- Without `lc:forKind`, the chain behind o13 has three ends: o0, oC and oG. +- New demand declares two entry acts. Its chain reaches only o0. No dependency leads from oC towards o16 for new demand, though `f:what-is-wrong` says such an item is "classified new at oC, then an option at o0". diff --git a/meta/sessions/2026-11-17-act-map-and-process/successor-items.md b/meta/sessions/2026-11-17-act-map-and-process/successor-items.md new file mode 100644 index 0000000..8b7ea9e --- /dev/null +++ b/meta/sessions/2026-11-17-act-map-and-process/successor-items.md @@ -0,0 +1,15 @@ +# Successor items — the act map and a process, 2026-11-17 + +Nine items stay open after `DDD-dec-122`. None is ruled. O-6 and O-7 become new issues (R-7, point 6). + +| # | Item | Where it shows | For | +|---|---|---|---| +| O-1 | `DDD-address-02`'s region says the act map is "a tree of occurrences". `DDD-address-06` says its nodes are act sites and the tree is containment. The region is superseded, annotated or left | `core/claims/DDD-address-02.yaml:8`; `method/01-map-and-index-acts.md:62` | Emil | +| O-2 | The second top act, changes to the SDLC itself, has no graph data. One sentence about it contradicts `DDD-ground-06`: "the same addresses with a different layer path" | `applications/sdlc/sdlc.md:451`, `:455` | Emil | +| O-3 | Four places call the SDLC's part of the lifecycle map "the act map" | `survey.md`, S-10 | A prose pass, after PR #73 | +| O-4 | A later o5 reads an earlier o6, and o6 reads o5. The data leaves the first edge out | `applications/sdlc/sdlc.md:106`; `applications/sdlc/acts/acts.ttl:88` | Emil | +| O-5 | New demand declares oC as an entry act. No dependency leads from oC towards o16 for new demand | `applications/pdlc/graph/demand-flow.ttl`; `rulings.md`, evidence | Emil | +| O-6 | **New issue.** The lifecycle has no index of determinables. Its six sources of pull are not marked taken up or not, and carry no rate of change | `applications/pdlc/graph/demand-flow.ttl` | Emil opens the issue | +| O-7 | **New issue.** Which method step builds the index: step 1, or step 3, which is not filed. The rate of change has no unit and no measure, and step 4's tick rate is not filed either | `method/01-map-and-index-acts.md`; `method/03-surface-decisions.md` | Emil opens the issue | +| O-8 | PR #73 rewrites prose under `applications/`. The survey's line numbers are `main`'s at `84aa466` | `survey.md` | Whoever lands second | +| O-9 | Carry R-2, R-4 and R-5 to Mantle's planning project. Mantle's ruling of 2026-11-17 is a view with a bound and no north | `DDD-address-08`, notes | Emil | diff --git a/meta/sessions/2026-11-17-act-map-and-process/survey.md b/meta/sessions/2026-11-17-act-map-and-process/survey.md new file mode 100644 index 0000000..7c279ef --- /dev/null +++ b/meta/sessions/2026-11-17-act-map-and-process/survey.md @@ -0,0 +1,156 @@ +# Survey — every use of "act map" + +"Act map" stands on 109 lines in 38 files. The repository uses the word in three ways that do not agree, and it defines none of them. + +| Question | What the uses say | +|---|---| +| One or many? | **One per mapped case.** No line has two maps of one case. The repository holds three cases, each with its own map: the dry run's T2a, the quick start's Varve slice, and the product lifecycle. Four lines name the SDLC's part of the lifecycle map as "the SDLC's act map", against seven that say the two files are one map | +| Tree or graph? | **Both, over different relations.** Containment is a tree: 11 lines, and the data enforces it. Dependency is a graph: 13 lines, and the data has acts with two and three incoming dependencies. Ten more lines read the map as a table with rows | +| Nodes? | **The claim says occurrences. The data says act sites.** `DDD-address-02` and step 1's card say "a tree of occurrences". `lc:Act` is "an act site of the map: one address, possibly many occurrences", and occurrences live in a separate record | + +Read at `84aa466` (`main`), with canon at `v5.15.2`. The search was `grep -rnIiE "act[- ]maps?\b|actmap|act_map"` over the working tree, excluding `.git` and this session's directory. + +## Findings + +| # | Finding | Where | +|---|---|---| +| S-1 | **Only one filed sentence says what the act map is, and it is in a region, not a statement.** `DDD-address-02`'s region says the claim "governs the act map (a tree of occurrences)". Step 1's card repeats it. No claim, decision or term defines the act map | `core/claims/DDD-address-02.yaml:8`; `method/01-map-and-index-acts.md:62` | +| S-2 | **Containment is a tree in the data.** Every act sits within exactly one composite act (`lc:ActShape`, `lc:within`, at most one). The lifecycle map has one root, `m:PDLC` | `applications/pdlc/graph/shapes.ttl:29`; `applications/pdlc/graph/acts.ttl:17` | +| S-3 | **Dependency is not a tree in the data.** o1b has three incoming dependencies, scoped by kind. o2b and o12 have two each. `lc:readsEarlier` adds loops across issues: o2a reads earlier o13s | `applications/sdlc/acts/acts.ttl`; `applications/pdlc/graph/ontology.ttl:113` | +| S-4 | **The stream has a loop between act sites that the data leaves out.** The prose has o5 read "o4 (then o6)" and o6 read o5's report. The data gives o5 a dependency on o4 only. The loop is between occurrences: a later o5 reads an earlier o6. "Within an issue, dependencies are acyclic" therefore holds over occurrences, and over sites only because the edge is omitted | `applications/sdlc/sdlc.md:95`, `:106`, `:107`; `applications/sdlc/acts/acts.ttl:88` | +| S-5 | **Nodes are act sites in the data and occurrences in the claim.** `lc:Act` is a site with many occurrences. The occurrence record is a second artefact, merged with the map for checking. The quick start's rows are called occurrences and then used as sites ("if you run o3 again tomorrow, the new occurrence has o3's address") | `applications/pdlc/graph/ontology.ttl:33`; `applications/sdlc/acts/check.py:10`; `method/quick-start.md:77`, `:101` | +| S-6 | **Each case's map has its own declared ground axes.** T2a declares four. The quick start declares three. The lifecycle declares three (`DDD-dec-115`: "the act map's three axes"). No map uses two declarations, and no declaration serves two maps | `meta/sessions/2026-09-27-act-address/results-act-address-dry-run.md:31`; `method/quick-start.md:63`; `core/decisions/DDD-dec-115.yaml:5` | +| S-7 | **One map already holds several roots.** T2a's layer paths start at three different composite acts: delivery pipeline, operations, runtime. Emil accepted that map (act address landing, GATE 1 ruling 7; `DDD-dec-38`) | `meta/sessions/2026-09-27-act-address/results-act-address-dry-run.md:35` to `:37` | +| S-8 | **"Top act" is used and not filed.** `sdlc.md` names "the second top act: changes to the SDLC itself", and has o11 read changes "produced in other top acts". The second top act has no graph data: `m:PDLC` is the only root. Nothing says whether it belongs to the lifecycle map or to a second map | `applications/sdlc/sdlc.md:124`, `:451` | +| S-9 | **A sentence in `sdlc.md` §10 contradicts `DDD-ground-06`.** "Its acts sit at the same addresses with a different layer path." The layer path is part of the address, so two acts with different layer paths never share an address. They can share a ground position, which is coincidence (`DDD-address-04`). Flagged, not harmonised (`AGENTS.md`) | `applications/sdlc/sdlc.md:455` | +| S-10 | **A part is named as a map in four places.** "The SDLC's act map", "its act map", "the act map, o1b to o13". Each sits beside a sentence saying the two files are one map | `applications/sdlc/acts/acts.ttl:5`; `applications/sdlc/README.md:68`; `applications/sdlc/sdlc.md:89` | +| S-11 | **"Process" is used and not filed.** Canon at `v5.15.2` has no term for it. Downstream, Emil's Round 9 sentence "the SDLC is a subprocess under the PDLC" was ruled to read "the SDLC is a composite act within the PDLC". Step 1b uses the word 24 times without saying what one is, and its four claims wait on "a second process" | `core/decisions/DDD-dec-114.yaml:18`; `method/01b-map-the-flow-of-demand.md`; `core/claims/DDD-flow-01.yaml:73` | +| S-12 | **The SDLC holds no act directly.** Its members are four composite acts: frame, build, accept, land and run. Every SDLC act sits within one of those. The PDLC holds eight acts directly, and the SDLC | `applications/sdlc/acts/acts.ttl:126` to `:130` | + +## The list + +"One per mapped case" marks a line that names a map belonging to one case or one run of step 1. "One" marks a line that says "the act map" inside one case. "Not implied" means the line names the map and says nothing of its shape. + +### In force: claims, decisions, the method, the applications + +| File and line | One or many | Tree or graph | What the line says | +|---|---|---|---| +| `applications/README.md:7` | one | tree: containment | nesting is "a relation in the act map" | +| `applications/pdlc/README.md:21` | one | not implied | "the PDLC's acts on the act map": the PDLC has acts on it, not a map of its own | +| `applications/pdlc/README.md:29` | one | tree: containment | nesting is a relation in the map | +| `applications/pdlc/graph/acts.ttl:5` | one, held in two files | not implied | "the PDLC's acts on the act map"; "the two files are one map" | +| `applications/pdlc/graph/check.py:1` | one, held in two files | not implied | "the act map (PDLC and SDLC)" | +| `applications/pdlc/graph/check.py:138` | one | graph: dependency edges | "follows an edge of the act map" | +| `applications/pdlc/graph/check.py:155` | one | table: rows | "a seam is a row of the act map" | +| `applications/pdlc/graph/demand-flow.ttl:11` | one, held in two files | not implied | names both acts.ttl files | +| `applications/pdlc/graph/ontology.ttl:11` | one | tree and graph | the form: composite acts, act sites, causal dependencies | +| `applications/pdlc/graph/ontology.ttl:33` | one | not implied | "an act site of the map": the nodes are sites, not occurrences | +| `applications/pdlc/graph/ontology.ttl:54` | one | not implied | dct:source names "the act map" of Round 2 as where a term comes from | +| `applications/pdlc/graph/ontology.ttl:62` | one | not implied | dct:source names "the act map" of Round 2 as where a term comes from | +| `applications/pdlc/graph/ontology.ttl:64` | one | not implied | dct:source names "the act map" of Round 2 as where a term comes from | +| `applications/pdlc/graph/ontology.ttl:66` | one | not implied | dct:source names "the act map" of Round 2 as where a term comes from | +| `applications/pdlc/graph/ontology.ttl:69` | one | not implied | dct:source names "the act map" of Round 2 as where a term comes from | +| `applications/pdlc/graph/ontology.ttl:70` | one | not implied | dct:source names "the act map" of Round 2 as where a term comes from | +| `applications/pdlc/graph/ontology.ttl:71` | one | not implied | dct:source names "the act map" of Round 2 as where a term comes from | +| `applications/pdlc/graph/ontology.ttl:72` | one | not implied | dct:source names "the act map" of Round 2 as where a term comes from | +| `applications/pdlc/graph/ontology.ttl:73` | one | not implied | dct:source names "the act map" of Round 2 as where a term comes from | +| `applications/pdlc/graph/ontology.ttl:92` | one | not implied | dct:source names "the act map" of Round 2 as where a term comes from | +| `applications/pdlc/graph/ontology.ttl:93` | one | not implied | dct:source names "the act map" of Round 2 as where a term comes from | +| `applications/pdlc/graph/ontology.ttl:94` | one | not implied | dct:source names "the act map" of Round 2 as where a term comes from | +| `applications/pdlc/graph/ontology.ttl:106` | one | not implied | dct:source names "the act map" of Round 2 as where a term comes from | +| `applications/pdlc/graph/ontology.ttl:114` | one | not implied | dct:source names "the act map" of Round 2 as where a term comes from | +| `applications/pdlc/graph/ontology.ttl:124` | one | not implied | dct:source names "the act map" of Round 2 as where a term comes from | +| `applications/pdlc/graph/ontology.ttl:125` | one | not implied | dct:source names "the act map" of Round 2 as where a term comes from | +| `applications/pdlc/graph/ontology.ttl:195` | one | not implied | dct:source names "the act map" of Round 2 as where a term comes from | +| `applications/pdlc/graph/shapes.ttl:8` | one | not implied | | +| `applications/pdlc/graph/shapes.ttl:252` | one | not implied | | +| `applications/pdlc/graph/shapes.ttl:367` | one | graph: dependency edges | an edge is a dependency or an earlier reading | +| `applications/pdlc/graph/shapes.ttl:393` | one | table: rows | "a seam is a row of the act map"; the row is a producing act and a reading act joined by a dependency | +| `applications/pdlc/pdlc.md:152` | one | table: rows | the seam table is checked against it | +| `applications/pdlc/pdlc.md:156` | one, held in two files | not implied | names both acts.ttl files | +| `applications/pdlc/pdlc.md:566` | one | not implied | "the PDLC's acts on the act map" | +| `applications/pdlc/pdlc.md:568` | one, held in two files | not implied | "one map, validated together"; "the form of method step 1" | +| `applications/sdlc/README.md:68` | a part named as a map | not implied | "its act map": the SDLC's own | +| `applications/sdlc/acts/acts.ttl:5` | a part named as a map | not implied | "the SDLC's act map"; four lines later "the two files are one map" | +| `applications/sdlc/acts/acts.ttl:13` | one | tree: containment | "the act map's layers, read as composite acts" | +| `applications/sdlc/acts/check.py:1` | one | not implied | "the lifecycle act map" | +| `applications/sdlc/acts/check.py:6` | one | not implied | | +| `applications/sdlc/acts/check.py:10` | one | not implied | the occurrence record is merged with the map: the map holds sites, not occurrences | +| `applications/sdlc/acts/shapes.ttl:16` | one | not implied | | +| `applications/sdlc/sdlc.md:89` | a part named as a map | not implied | heading "The act map, o1b to o13"; the body says "one map" | +| `applications/sdlc/sdlc.md:118` | one | graph: dependency edges | readings from the causal index | +| `applications/sdlc/sdlc.md:665` | one | graph: dependency edges | dependencies scoped by kind | +| `applications/sdlc/sdlc.md:762` | one | not implied | another product's sessions would be "mapped onto the act map": one map that adopters share | +| `core/claims/DDD-address-02.yaml:8` | one | tree: containment | "a tree of occurrences"; the only claim that says what the map is | +| `core/claims/DDD-flow-03.yaml:14` | one, held in two files | not implied | names both acts.ttl files | +| `core/claims/DDD-flow-03.yaml:22` | one | tree and graph | traced on the map; the seam table has rows read at an act | +| `core/decisions/DDD-dec-114.yaml:20` | one | tree: containment | nesting is "a relation in the act map" | +| `core/decisions/DDD-dec-114.yaml:128` | one | table: rows | a seam listing is checked against it | +| `core/decisions/DDD-dec-115.yaml:5` | one | not implied | the map has three axes: one axis declaration | +| `core/decisions/DDD-dec-115.yaml:124` | one | not implied | "the act map's graph data" | +| `core/decisions/DDD-dec-115.yaml:185` | one | graph: dependency edges | readings from the causal index | +| `core/decisions/DDD-dec-38.yaml:55` | one per mapped case | not implied | "the T2a act map" | +| `method/01-map-and-index-acts.md:62` | one | tree: containment | "a tree of occurrences" | +| `method/01-map-and-index-acts.md:82` | one per mapped case | not implied | "this step's act map" | +| `method/01b-map-the-flow-of-demand.md:24` | one | graph: dependency edges | demand is traced across it | +| `method/01b-map-the-flow-of-demand.md:41` | one per mapped case | not implied | step 1's output | +| `method/01b-map-the-flow-of-demand.md:57` | one | graph: dependency edges | "along edges of the act map" | +| `method/01b-map-the-flow-of-demand.md:97` | one per mapped case | not implied | "your act map" | +| `method/01b-map-the-flow-of-demand.md:132` | one per mapped case | not implied | ground in | +| `method/01b-map-the-flow-of-demand.md:134` | one | not implied | graph data sits beside it | +| `method/01b-map-the-flow-of-demand.md:145` | one, held in two files | not implied | names both acts.ttl files | +| `method/01b-map-the-flow-of-demand.md:249` | one per mapped case | not implied | | +| `method/quick-start.md:75` | one per mapped case | table: rows | a table, one row per occurrence, with the layer path as a column | + +### Release notes + +| File and line | One or many | Tree or graph | What the line says | +|---|---|---|---| +| `CHANGELOG.md:15` | one per mapped case | graph: dependency edges | demand is traced across it | +| `CHANGELOG.md:25` | one per mapped case | not implied | each specification carries its own | +| `releases/v0.7.0.yaml:41` | one per mapped case | graph: dependency edges | copy of CHANGELOG.md:15 | +| `releases/v0.7.0.yaml:51` | one per mapped case | not implied | copy of CHANGELOG.md:25 | + +### Session records (history, never edited) + +| File and line | One or many | Tree or graph | What the line says | +|---|---|---|---| +| `meta/sessions/2026-09-27-act-address-landing/gate1-rulings.md:15` | one per mapped case | not implied | the T2a map is accepted | +| `meta/sessions/2026-09-27-act-address-landing/gate1-survey.md:145` | one per mapped case | not implied | | +| `meta/sessions/2026-09-27-act-address-landing/gate1-survey.md:148` | one per mapped case | not implied | axes and five occurrences | +| `meta/sessions/2026-09-27-act-address-landing/gate1-survey.md:174` | one per mapped case | not implied | | +| `meta/sessions/2026-09-27-act-address-landing/prompt.md:48` | one per mapped case | not implied | | +| `meta/sessions/2026-09-27-act-address/prereg-act-coordinates.md:26` | one | tree: containment | R-6: "a tree of occurrences" | +| `meta/sessions/2026-09-27-act-address/prereg-act-coordinates.md:77` | one | tree: containment | "a tree of unique occurrences" | +| `meta/sessions/2026-09-27-act-address/prereg-act-coordinates.md:108` | one per mapped case | not implied | "its act map": the first independent case has its own | +| `meta/sessions/2026-09-27-act-address/results-act-address-dry-run.md:7` | many (plural) | not implied | "the act maps below": one per case | +| `meta/sessions/2026-09-27-act-address/results-act-address-dry-run.md:11` | one per mapped case | not implied | the map depends on the axes the mapper declared | +| `meta/sessions/2026-09-27-act-address/results-act-address-dry-run.md:31` | one per mapped case | table: rows | a table of occurrences under its own four declared axes; its layer paths start at three different roots | +| `meta/sessions/2026-09-27-act-address/results-act-address-dry-run.md:70` | one per mapped case | not implied | | +| `meta/sessions/2026-09-27-act-address/rulings.md:27` | one per mapped case | not implied | "the T2a act map's axis declarations" | +| `meta/sessions/2026-11-02-ready-to-try/predictions.md:378` | one per mapped case | not implied | the quick start's map | +| `meta/sessions/2026-11-02-ready-to-try/predictions.md:421` | one per mapped case | table: rows | the quick start's map, counted in rows | +| `meta/sessions/2026-11-07-sdlc-and-pdlc/findings.md:10` | one | graph: dependency edges | an order between two acts | +| `meta/sessions/2026-11-07-sdlc-and-pdlc/findings.md:19` | one | graph: dependency edges | causal-index readings | +| `meta/sessions/2026-11-07-sdlc-and-pdlc/predictions.md:57` | one | not implied | a record is merged with the map: the map holds sites, not occurrences | +| `meta/sessions/2026-11-07-sdlc-and-pdlc/prompt-2.md:208` | one | table: rows | a row is checked against the map | +| `meta/sessions/2026-11-07-sdlc-and-pdlc/prompt-2.md:240` | one | not implied | "the act map, the occurrence record": two things | +| `meta/sessions/2026-11-07-sdlc-and-pdlc/prompt.md:35` | one | not implied | | +| `meta/sessions/2026-11-07-sdlc-and-pdlc/prompt.md:197` | one | tree: containment | nesting is "a relation in the act map" | +| `meta/sessions/2026-11-07-sdlc-and-pdlc/prompt.md:314` | one | not implied | the o1 to o15 map | +| `meta/sessions/2026-11-07-sdlc-and-pdlc/prompt.md:373` | one | not implied | "the act map's form" | +| `meta/sessions/2026-11-07-sdlc-and-pdlc/prompt.md:438` | one | table: rows | seam rows are checked against it | +| `meta/sessions/2026-11-07-sdlc-and-pdlc/prompt.md:478` | one | not implied | "the PDLC's acts on the act map" | +| `meta/sessions/2026-11-07-sdlc-and-pdlc/prompt.md:498` | one | not implied | | +| `meta/sessions/2026-11-07-sdlc-and-pdlc/prompt.md:504` | a part named as a map | not implied | "the act map for o1b to o13" | +| `meta/sessions/2026-11-07-sdlc-and-pdlc/rulings.md:39` | one | not implied | | +| `meta/sessions/2026-11-07-sdlc-and-pdlc/rulings.md:201` | one | tree: containment | nesting is "a relation in the act map" | +| `meta/sessions/2026-11-07-sdlc-and-pdlc/rulings.md:576` | one | graph: dependency edges | causal-index readings | +| `meta/sessions/2026-11-07-sdlc-and-pdlc/rulings.md:710` | one | tree: containment | R-9, the ruled reading | +| `meta/sessions/2026-11-07-sdlc-and-pdlc/rulings.md:717` | one | graph: dependency edges | causal-index readings | +| `meta/sessions/2026-11-07-sdlc-and-pdlc/rulings.md:719` | one | table: rows | the seam table is checked against it | +| `meta/sessions/2026-11-07-sdlc-and-pdlc/rulings.md:739` | one | not implied | the o1 to o15 map, verbatim | +| `meta/sessions/2026-11-08-demand-flow/findings.md:41` | one | not implied | | +| `meta/sessions/2026-11-08-demand-flow/successor-items.md:14` | one | not implied | | +| `meta/sessions/README.md:53` | one | not implied | "the lc: act map" | +| `meta/sessions/README.md:55` | one | not implied | | diff --git a/meta/sessions/2026-11-18-act-map-scope-and-pull/bootstrap.md b/meta/sessions/2026-11-18-act-map-scope-and-pull/bootstrap.md new file mode 100644 index 0000000..a71de08 --- /dev/null +++ b/meta/sessions/2026-11-18-act-map-scope-and-pull/bootstrap.md @@ -0,0 +1,56 @@ +# Bootstrap — the act map, scope and pull, 2026-11-18 + +Filled in at arrival, 2026-11-18 at 08:51 UTC (09:51 Europe/Berlin), and committed with `prompt.md` as the session's first commit. + +| Parameter | Value | +|---|---| +| Invocation | A Claude Code session started by Emil. `prompt.md` is the prompt, verbatim | +| Principal | Emil. He rules each proposal at G1 and G2, and merging is his | +| Issue | [#80](https://github.com/mindovermachine-dev/decision-driven-design/issues/80), the charter, read in full at arrival. It had no comments. Every commit of this session carries `Refs #80` on its own line | +| Downstream base | `84aa466`, head of `main` at arrival | +| Downstream branch | `claude/act-map-scope-and-pull-2026-11-18`, from `main` at `84aa466` | +| Archive branch | `claude/pull-and-scope-2026-11-18`, head `47cf4cc`, as the prompt says. A source to copy from. Not merged, not rebased, not pushed to, not deleted | +| Old branch | `claude/act-map-and-process-2026-11-17`, head `66bc21f`, still on the remote. Every commit on it is contained in the archive branch (checked with `git merge-base --is-ancestor`). Its deletion is Emil's (2026-11-18 R-7), not this session's | +| Upstream repo | github.com/mindovermachine-dev/actor-indexed-determination, read only | +| Upstream pin | `v5.15.2` in `graph/upstream.yaml`, tag at `084bce7`. Upstream `main` is also at `084bce7`. Unchanged by this session | +| Upstream fetched | Cloned at arrival with every branch and tag. Canon is read at `v5.15.2` only | + +## Gates + +| Gate | What | Who | +|---|---|---| +| G0 | `prompt.md` and this file committed before any other change | Claude | +| D-1 | The two session directories copied from `47cf4cc` as they stand, in one commit naming the source commit | Claude | +| G1 | Deliverables 2 to 5: the pull statement and the five-step procedure with one worked example tested against a junior engineer's first contact; the proposed claims; what each of steps 2 to 10 adds to the act map; the effect on T2b's pre-registration, on step 1b's kinds and on `DDD-address-02`'s region. **Stop for Emil**, who rules each proposal and where it is filed | Claude; Emil rules | +| G2 | Deliverable 6, on the G1 rulings only: claims filed, cards changed by supersession. The whole `validate` workflow run locally before every push, and the workflow run on each pushed commit reported. A pull request against `main`, naming the head to ratify. **Stop for Emil** | Claude; Emil rules | +| Merge | Landing the branch | Emil | + +## Decision and claim ids + +Checked at arrival across every branch of both repositories: 63 remote branches downstream and 28 upstream, with every upstream tag. + +| Id | Where it appears | +|---|---| +| `DDD-dec-122` | Files on the archive branch and the old branch only. Nowhere on `main` | +| `DDD-dec-121` | `claude/confident-goldberg-c0y0ia` | +| `DDD-dec-119`, `-120` | `claude/zealous-ptolemy-3txf8k` | +| Highest on `main` | `DDD-dec-118` | +| Highest upstream | `DDD-dec-39`, on any branch or tag | +| `DDD-address-06` to `-08` | Files on the archive branch and the old branch only | +| `DDD-address-09` onwards | No file on any branch of either repository. The only mention is the text of 2026-11-18 R-8 on the archive branch | +| `DDD-address-*` upstream | None | + +So `DDD-dec-122` and `DDD-address-06` to `-08` are this issue's by 2026-11-18 R-8, and new claims start at `DDD-address-09`. A new decision, if one is ruled, would take `DDD-dec-123`. The check is run again before filing. + +## Deviations + +| # | Deviation | Why | +|---|---|---| +| V-1 | The session environment named a different development branch. The branch is the one the prompt names | The prompt is Emil's explicit instruction | +| V-2 | The model is not recorded here, unlike earlier bootstraps | The session's operating rules keep model identifiers out of files pushed to the repository | +| V-3 | `predictions.md` is not in this first commit | The prompt says to commit `prompt.md` and this file first. Predictions are written before each later operation | + +## Known before work starts + +- The `Agent review` check fails on `main` on a build error in `tools/canon` (`NU1004`: the lock file expects `Microsoft.NET.ILLink.Tasks` 10.0.14 and the runner resolves 10.0.15). This branch does not cause it and does not fix it. It is reported at each push. +- The local toolchain has `dotnet` 10.0.404 and `python3`, so the whole `validate` workflow can be run locally. diff --git a/meta/sessions/2026-11-18-act-map-scope-and-pull/effects.md b/meta/sessions/2026-11-18-act-map-scope-and-pull/effects.md new file mode 100644 index 0000000..9e81f43 --- /dev/null +++ b/meta/sessions/2026-11-18-act-map-scope-and-pull/effects.md @@ -0,0 +1,75 @@ +# Effects on T2b, on step 1b's kinds and on `DDD-address-02`'s region + +Stated before any card changes, as #80 requires. Nothing below is applied. Each effect is `[PROPOSED]` until Emil rules Q-10 in `proposals.md`. + +| Object | Stands | Superseded, if the proposals are ruled | Registered again | +|---|---|---|---| +| T2b's pre-registration | All of it, as text | One sentence, in its reading only | Six items for the T2b pre-registration that is not yet sealed | +| Step 1b's kinds | The PDLC's three kinds and `DDD-flow-01` to `-04` | The subject that declares kinds, and classification by source | The promotion item, with two new columns | +| `DDD-address-02` | Its statement, falsifier, test and breaks | "(a tree of occurrences)" in its region | Nothing | + +## T2b's pre-registration + +**Weakest point.** The pre-registration says T2b's "act map is itself produced by method step 1". If step 1 gains a first part (S-2, option A), T2b's map is produced by a step that did not exist when the predictions were fixed. The predictions survive, because they are about the map and not the procedure. But the case is then also the first test of the pull procedure, which nobody pre-registered. + +**What stands.** `meta/sessions/2026-09-27-act-address/prereg-act-coordinates.md`, revision 3, is fixed and is never edited. Its criteria (§2), its predictions (§5) and its falsifiers (§6) concern the address, the layer path, the time index and reuse. None concerns scope, pulls, the index of determinables or the four states. Every one stands as written. The T2b test items accumulated at the act address landing (`gate2-rulings.md`: F-1 and verdict-triggered decisions) and the deferred What falsifier (`2026-10-28-product-framework-what-move/falsifier-t2b.md`) stand. + +**What is superseded, in reading only.** Its R-6 row says: "The act map is therefore a tree of occurrences". Under `DDD-address-06` the map's nodes are act sites, the tree is containment, and occurrences are recorded beside the map. The sentence is not edited. A reader of the pre-registration reads it under `DDD-address-06`, as a reader of a published paper reads it under a later erratum. + +**What is registered again.** The T2b pre-registration proper, `prereg-t2b.md`, is not yet written or sealed: its design session stopped at its Gate 0 (`claude/t2b-design-2026-10-26`, head `50807e0`). These items are recorded for it, in the form `gate2-rulings.md` used. Writing and sealing it is that session's, not this one's. + +| Item | Source | What T2b must contain or check | +|---|---|---| +| T-1 | `DDD-address-06`, P-1 | Its map is built in the five parts: scope, act sites, layers, dependencies and the index. Each falsifier limb is searched for, the fourth among them: a composite act that straddles the scope | +| T-2 | `DDD-address-09`, P-4 | The five-part procedure is run on the slice's scope before any act is mapped. Every field of every pull statement passes or fails its check, and the result is recorded per field | +| T-3 | `DDD-address-09`, P-4 | If two practitioners are available: both run the procedure without conferring, and their entry acts are compared | +| T-4 | `DDD-address-10`, P-5 | The axis kinds are read on is declared before the first kind is read. No kind is assigned by its source | +| T-5 | `DDD-address-11`, P-6 | "Actual" is never filled. Any demand item whose gap needs a fifth state is recorded as a finding | +| T-6 | The pre-registration's own §4 | The pre-registration states which version of step 1's card produced the map, by commit | + +## Step 1b's kinds + +**Weakest point.** Step 1b's worked example, its table of sources and its field-report form are all organised by source of pull. Under `DDD-address-10` the source no longer decides the kind. So the parts of the card a reader meets first are the parts this changes. + +**What stands.** +- The PDLC's three kinds: new, owed and own-ground, and their one-line tests. +- `DDD-flow-01` to `-04`, projected and scoped to the PDLC. `DDD-flow-01` already tests each kind relative to the decisions in force, which is a reading of ground, so 11-18 R-4 confirms it rather than contradicts it. +- `DDD-dec-118`'s place for the step, after step 1, and its ruling that the PDLC's kinds are a starting point and not the method's fixed set. +- "Sighting is not entry" (`DDD-flow-03`, Round 26). + +**What is superseded, if ruled.** + +| Card text | Becomes | Basis | +|---|---|---| +| "each process declares its own kinds of demand" | Each map declares its kinds, as positions on a declared ground axis | P-3: a process is not an object of the model, so it cannot declare. P-5: a kind is a position | +| Part 6, "Classify each source of pull into the kinds of demand the process declares" | Read each pull's kind from its position. The source is kept as "sighted in", and decides nothing | P-5; 11-18 R-4 | +| Part 1, "Map acts forward from the request" | Map acts forward from the entry acts of the scope's pull statements | S-2, if ruled | + +These are made by a superseding section on the card, and the lines stay as written. + +**What is registered again.** The promotion item P-1 of the demand-flow session asks whether a second process declares the PDLC's kinds. It becomes: whether a second map's declared axis and positions match the PDLC's. The field-report table gains two columns: the axis, and the position. W-3 of `pull-procedure.md`, own-ground demand as a position on the arrangement's determinables, is registered there as an open item, not as a ruling. + +## `DDD-address-02`'s region + +**Weakest point.** Every option leaves a filed file saying something another filed file contradicts, or changes a filed file. There is no way to correct a region alone without one of the two. + +**What stands.** The statement, "Every act occurrence is its own act, and reuse of code across occurrences is a derivation-layer relation — many occurrences pulling one component — never an identity between acts". Its falsifier, its test and its breaks. Every occurrence is still its own act under `DDD-address-06`. + +**What is superseded.** Its region's "Governs the act map (a tree of occurrences)". Under `DDD-address-06` the act map's nodes are act sites, and occurrences are recorded beside it. + +**Three options, for Emil.** + +| | Option 1 (recommended) | Option 2 | Option 3 | +|---|---|---|---| +| What is done | A `notes` entry is appended to `DDD-address-02`, saying that "(a tree of occurrences)" is superseded by `DDD-address-06` under `DDD-dec-122`. The region is not reworded | `DDD-address-02` is retired, and a successor with the same statement and a corrected region is filed with `supersedes: DDD-address-02` | Nothing. `DDD-address-06`'s notes flag the difference, as on 2026-11-17 (O-1) | +| For it | Nothing is rewritten. The claim file carries its own correction, as `AGENTS.md` asks of prose that contradicts a claim | The region and the canon agree | No filed file changes | +| Against it | It corrects a region by a note, which the claim format has no field for | Retired means "overturned or shown not to follow", and the statement was not. It also moves every citation of `-02` | Two projected claims in one area disagree, and a reader of `-02` alone is misled | + +Step 1's card repeats the phrase in its "Artefact out" row. Under any option it is changed by a superseding section on the card. + +## What changes nothing + +- `DDD-address-01`, `-03`, `-04` and `-05`. +- Every act and every piece of lifecycle data under `applications/` (out of scope by #80). +- Canon upstream, and the pin. +- The registers of the step cards. Step 1 stays ruled for test and step 1b stays `[PROPOSED]`, whichever option is ruled for S-2. diff --git a/meta/sessions/2026-11-18-act-map-scope-and-pull/predictions.md b/meta/sessions/2026-11-18-act-map-scope-and-pull/predictions.md new file mode 100644 index 0000000..48353f6 --- /dev/null +++ b/meta/sessions/2026-11-18-act-map-scope-and-pull/predictions.md @@ -0,0 +1,67 @@ +# Predictions + +Each prediction is written and committed before its operation runs. The result is recorded against it afterwards, and a miss is kept as a miss. + +## Operation 1 — copy the two session directories from `47cf4cc` (deliverable 1) + +| # | Prediction | Result | +|---|---|---| +| P-1.1 | Eleven files are added: eight under `2026-11-17-act-map-and-process/` and three under `2026-11-18-pull-and-scope/`. No other path changes | Held. Eleven files added, eight and three, in commit `e1e2491`; no other path | +| P-1.2 | Every copied file is byte-identical to its blob at `47cf4cc` (`git diff 47cf4cc HEAD --` over the two directories is empty) | Held. `git diff --cached 47cf4cc` over the two directories was empty before the commit | +| P-1.3 | `survey.md` is among them and says it covers 109 lines in 38 files read at `84aa466` | Held. Its line 11 reads "Read at `84aa466` (`main`)"; its opening says 109 lines in 38 files | +| P-1.4 | No file under `core/`, `method/` or `meta/sessions/README.md` changes | Held | +| P-1.5 | `python3 scripts/ci.py` passes after the copy, as it does on `main` | Held, after a local setup step. The first run failed the `what` and `how` gates because `rdflib`, `pyoxigraph` and `pyld` were not installed in the container. Installed from `scripts/requirements.txt`; the second run passed every gate. Not a branch fault | + +## Operation 2 — draft deliverables 2 to 5 for G1 + +Written before any of the four drafts. Nothing under `core/`, `method/` or `applications/` changes in this operation. + +| # | Prediction | Result | +|---|---|---| +| P-2.1 | Every field of the pull statement gets a check that passes or fails on the worked example. At least one check is a judgement, not mechanical || Held. Thirteen fields, each with a check. F-4 and F-12 are judgement, and F-10 is judgement in declaring its axis | +| P-2.2 | With the How as the scope, the owed item's trace stops at the edge: the classification is outside the scope, so the entry act inside the scope is not the entry act `DDD-flow-03` names for the PDLC || Held. Under the How, #103 enters at plan the realisation; under the PDLC, at the classification (`pull-procedure.md`, W-1) | +| P-2.3 | "Actual" cannot be filled for any determinable in the worked example || Held, for all three items | +| P-2.4 | The own-ground item fits no gap on the product's determinables. It fits only once a determinable of the arrangement itself is named || Held. #104 fits "expected, with nothing specified" only on the arrangement's determinable (W-3) | +| P-2.5 | At least one desk check against a junior engineer's first contact fails on vocabulary: a word the procedure uses without defining it || Held. Part 1 fails on the owner names. Part 4 also fails, without the axis: a failure the prediction did not foresee | +| P-2.6 | All nine "Value made visible" rows of steps 2 to 10 can be filled from the draft line each card already quotes. None has to say it cannot be. At least two steps attach beside the map, not in it || Held. All nine rows filled from the quoted line. Four steps attach beside the map (6, 7, 8, 9) | +| P-2.7 | Six claims are proposed: `DDD-address-06` to `-08` amended and `-09` to `-11` new. `DDD-dec-122` is amended and no new decision id is needed || Held | +| P-2.8 | The proposals surface at least one conflict between a ruling of 2026-11-17 and one of 2026-11-18 that only Emil can settle || Held. 11-17 R-7 point 3 against 11-18 R-2 and R-3, on what individuates a map (`proposals.md`, P-2) | +| P-2.9 | `python3 scripts/ci.py` passes after the drafts are committed || Held | + +## Operation 3 — record the G1 rulings and amend the session's proposals + +`rulings.md` was written before this section, on receipt of Emil's message, as he asked. The predictions below cover the amendments that follow it. Nothing under `core/` or `method/` changes in this operation, because three conflicts are held for Emil (`rulings.md`, K-1 to K-3). + +| # | Prediction | Result | +|---|---|---| +| P-3.1 | Every amendment is appended as a superseding section. No line written at G1 is reworded || Held. Each amendment is an appended section; `git diff` shows additions only in the three files | +| P-3.2 | Part 1 can be stated in plain words with no owner name, and the worked example still passes checks F-1 and F-2 || Held (`pull-procedure.md`, G1-A2) | +| P-3.3 | With "built or not built" as the method's offered axis, #101 and #103 are told apart, and #104 is not, without "whose determinable" || Held (G1-A3): #104 sits with #103 on "built" | +| P-3.4 | Adding "expected against observed" gives #104 a second reading, beside "expected, with nothing specified" || Held (G1-A4). #103 also gains it, from the customer's side, which the prediction did not foresee | +| P-3.5 | `python3 scripts/ci.py` passes || Held | + +## Operation 4 — file as ruled (deliverable 6) + +Written after Emil's answers to K-1 to K-3 (13:26) and before any file under `core/` or `method/` is written. The id check was run again at 12:26 UTC: 65 branches downstream, nothing at `DDD-address-09` or above, `DDD-dec-122` only on the archive and the old branch, and upstream's highest decision `DDD-dec-39`. + +| # | Prediction | Result | +|---|---|---| +| P-4.1 | Seven new files under `core/`: `DDD-address-06` to `-11` and `DDD-dec-122`. One filed file changes, `DDD-address-02`, by an appended `notes` entry only. No other line under `core/` changes || Held. `fe9d72f`: seven files added, `DDD-address-02` gains six `notes` lines and loses none | +| P-4.2 | `validate-claims.py` passes both trees with no new error. The single-limb warning fires on none of the six new statements || **Missed, in part.** No error. The single-limb warning fires on `DDD-address-11`, whose statement, ruled as written (G1-Q5), joins its four states with semicolons. Left as ruled | +| P-4.3 | `DDD-dec-122`'s `basedOn` resolves in full: every entry is a file here or a pin in `graph/upstream.yaml`. `DDD-ground-05`, `-06`, `-01` and `DDD-frame-13` are already pinned at `v5.15.2`, so no pin is added and `graph/upstream.yaml` is unchanged || Held. `validate-claims.py --decisions` passes, and `graph/upstream.yaml` is unchanged | +| P-4.4 | Eleven cards change, each by an appended superseding section: steps 1, 1b and 2 to 10. No existing line of any card is reworded || Held in substance. Eleven cards gain one section each, 0 lines removed. The sections are inserted before an existing heading, not appended at the end, so a reader meets them first | +| P-4.5 | Every commit carries `Refs #80`. Every commit that touches `core/` carries `Basis:`, and `canon issue-refs --basis` checks them and passes || Held, after a correction. The first local run reported "Refs checked on 0, Basis on 0": the container's clone was shallow, so the rule's start commits were not ancestors and nothing was covered. After `git fetch --unshallow`: "Refs checked on 10, Basis on 1", passing. CI fetches full history | +| P-4.6 | The whole `validate` workflow passes locally, and the workflow run on the pushed head succeeds || Held. Local: every step of `validate` passed on `3d76c88`. On GitHub, on `3d76c88`: `validate` passed on push (run 386) and on the pull request (run 387). `Agent review` failed (run 87) on `NU1004` in `tools/canon`, the known lock-file error #80 names, not on this branch. After #83 merged (main at `1bf7e56`), Emil asked for this result to be pushed so both required checks run again on the new head. Their results are reported to Emil in the session conversation and not written here, because writing them would move the head again | +| P-4.7 | No file under `applications/`, `tools/canon/` or the archive branch changes || Held. No path under `applications/`, `tools/` or `graph/` differs from `main`, and the archive branch was not pushed to | + +## Operation 5 — rebase onto the new `main` (`29cead8`) + +Written before the rebase, on Emil's instruction of 2026-11-19. The ids were checked again first (10:48 UTC): `DDD-dec-122` and `DDD-address-06` to `-11` are files only on this branch, the archive and the old branch. `DDD-dec-123` is on `main`. Upstream (27 branches and every tag) has no address claim, and its highest decision is `DDD-dec-39`. `main` mentions `DDD-address-06` once, as text, in `meta/sessions/2026-10-26-housekeeping/prompt.md:61`, as an option that was never filed. That line was on `main` at arrival. + +| # | Prediction | Result | +|---|---|---| +| P-5.1 | One commit conflicts, the one that adds this session's row to `meta/sessions/README.md`. Its only conflict is that row against the `2026-11-18-canon-lock-file/` row `main` added. `main` added no other row since `84aa466` || Held. One commit conflicted (the README row commit), on that row against `2026-11-18-canon-lock-file/`. `main` added no other row since `84aa466`: Emil's note that #85 and #86 also added rows does not match `main` at `29cead8`, where #86's session is the canon-lock-file row | +| P-5.2 | Resolved by keeping both rows in date order. Both are dated 2026-11-18, and this session arrived first (08:51 UTC, against 12:30 UTC), so its row goes first. No other line changes || Held. Both rows kept, this session's first. `git diff origin/main..HEAD -- meta/sessions/README.md` is one added line | +| P-5.3 | After the rebase, `git diff origin/main..HEAD` touches the same 40 paths as #84 did before, with the same content apart from the README row's position || Held. The same 40 paths. Against `0e76242`, the only content change outside the README row is this operation's section of this file | +| P-5.4 | Every rebased commit is signed and keeps its `Refs #80` line. `fe9d72f`'s successor keeps its `Basis:` line, and `issue-refs --basis` passes from `main` || Held. 14 commits, each signed (`G`) with `Refs #80`. `issue-refs --basis`: "14 commits in 29cead8673b3..; Refs checked on 14, Basis on 1" | +| P-5.5 | The whole `validate` workflow passes locally. The workflow results on the pushed head are reported to Emil and not written here, because writing them would move the head || Held locally: `ci.py` all gates passed; Canon.Tests 66 of 66; `changelog --check`, `issue-refs --basis` and `check --compare` passed. The workflow results on the pushed head are reported to Emil | diff --git a/meta/sessions/2026-11-18-act-map-scope-and-pull/prompt.md b/meta/sessions/2026-11-18-act-map-scope-and-pull/prompt.md new file mode 100644 index 0000000..2e95f50 --- /dev/null +++ b/meta/sessions/2026-11-18-act-map-scope-and-pull/prompt.md @@ -0,0 +1,82 @@ +Work issue #80 in mindovermachine-dev/decision-driven-design: +"The act map, scope and pull: what they are, and how step 1 starts from them". + +Read first, in this order +1. Issue #80, in full. It is the charter. Its "State at opening" section is part of it. +2. AGENTS.md and meta/way-of-working.md. +3. The two session records on the archive branch claude/pull-and-scope-2026-11-18 + (head 47cf4cc when this prompt was written; check it): + meta/sessions/2026-11-17-act-map-and-process/ and meta/sessions/2026-11-18-pull-and-scope/. + rulings.md in each is the ruled ground. proposals-2.md and survey.md in the first are material. +4. The ground files the issue names, and canon at the tag in graph/upstream.yaml. + Fetch upstream at that tag. Never cite an id or a canon sentence from memory. + +Set up before any other change +- Branch from main. Name it claude/act-map-scope-and-pull-. +- Create meta/sessions/-act-map-scope-and-pull/ with prompt.md (this prompt, verbatim) + and bootstrap.md (parameters, gates, base and archive heads, the decision-id check across + every branch of both repositories, deviations). Commit them first. +- Write predictions.md before each operation, and record the result against each prediction. + +The archive branch +- It is a source to copy from, never a branch to merge or rebase. Its commits carry no Refs line. +- Do not delete it and do not push to it. It keeps DDD-dec-122 and DDD-address-06 to -08 + visible as taken until this branch is pushed. +- Deliverable 1: copy the two session directories from it as they stand, in one commit that + names the source commit. Reword nothing in them. Do not copy its claims, its decision, its + two method sentences or its row in meta/sessions/README.md. Those are written by this session. + +Rulings already given (do not re-ask; they are in the two rulings.md files and in #80) +- Ids: the act map, a second act map, and a process keep DDD-address-06, -07, -08, and the + decision keeps DDD-dec-122, each amended before landing. New claims take DDD-address-09 + onwards. Each amended claim says in its notes that it was first drafted on 2026-11-17 on a + branch that did not land and was amended on the rulings of 2026-11-18. +- Everything else listed under "Ruled before this issue was opened" in #80. +"Stated, not ruled" in #80 is not ruled. Propose it; do not treat it as ground. + +Work the deliverables in the issue's order, and stop at the gates +- Deliverables 2 to 5, then STOP at G1. Give me each proposal with its weakest point first, + its falsifier, what it breaks, and which filed claim or ruling it rests on, or that it rests + on none. State the effect on T2b's pre-registration, on step 1b's kinds and on + DDD-address-02's region before any card changes. I rule each proposal and where it is filed. +- Deliverable 6 only after my G1 rulings. Change ruled material by supersession, never by + rewriting. Then STOP at G2. Open a pull request against main and name the head to ratify. + Merging is mine. + +Discipline +- Every claim is Ruled, [PROPOSED] or [OPEN]. You propose; I rule. Nothing self-ratifies. +- Mark reasoning I have not confirmed as "UNVERIFIED — Emil review". +- Projected means derived and unexercised. Reported means exercised. Do not fuse them. +- Record my rulings in the session's rulings.md when I give them, in my words, with the time. + Record corrections, including your own, with attribution. +- Carry one worked example through every part of the pull procedure, and test each part + against a junior engineer's first contact. Phoenix is never the worked example and is never + named. +- Human-read .md files start with the information itself, not with a description of the file. + British spelling. One idea per sentence. Tables for structure, prose for arguments. +- Use canon terms exactly as registered. Do not coin Danish terms. +- The four states are credited to GitHub user pjarnhus. Write the handle without an @. + Only the four names are his. The mapping to canon is ours, made from the names. + +The gate that must pass +- Run the whole validate workflow locally before every push, not only scripts/ci.py: + python3 scripts/ci.py + cd tools/canon + dotnet test --project tests/Canon.Tests -c Release + dotnet run --project src/Canon -c Release -- changelog --check + dotnet run --project src/Canon -c Release -- issue-refs --basis + dotnet run --project src/Canon -c Release -- check --compare +- Every commit carries "Refs #80" as its own line. Every commit that changes core/ also + carries a "Basis:" line naming the claim and decision ids it rests on, and names the ids it + touches. +- After each push, look at the workflow run for that commit and report its result. Do not + report a gate as passing from a local run alone. +- The "Agent review" check fails today on a build error in tools/canon (NU1004, a package + lock file mismatch). The branch does not cause it. Report it; do not fix it here. + +Out of scope +- Building the lifecycle's index of determinables. Changing any act or any lifecycle data + under applications/. Mantle. Canon in the principle repository. tools/canon. +- Cutting a release. + +If the ground for a ruling is missing, ask me. Do not fill the gap. diff --git a/meta/sessions/2026-11-18-act-map-scope-and-pull/proposals.md b/meta/sessions/2026-11-18-act-map-scope-and-pull/proposals.md new file mode 100644 index 0000000..aed8f17 --- /dev/null +++ b/meta/sessions/2026-11-18-act-map-scope-and-pull/proposals.md @@ -0,0 +1,201 @@ +# Proposals for G1 + +Six claims and one decision, written once with the rulings of both days in them. Three claims keep their ids and are amended before landing (2026-11-18 R-8): `DDD-address-06`, `-07` and `-08`. Three are new: `DDD-address-09` to `-11`. The decision keeps `DDD-dec-122`. Nothing here is filed. Every statement below is `[PROPOSED]` until Emil rules it. + +Each proposal gives its weakest point first, then the statement, the falsifier, what it breaks and what it rests on. "Rests on" names filed claims, rulings or nothing. Ruling numbers are by day: "11-17 R-7" is R-7 of `meta/sessions/2026-11-17-act-map-and-process/rulings.md`, and "11-18 R-2" is R-2 of `meta/sessions/2026-11-18-pull-and-scope/rulings.md`. + +| # | Id | What | Kind | Status proposed | +|---|---|---|---|---| +| P-1 | `DDD-address-06` | The act map, with its scope and its index | conceptual | projected | +| P-2 | `DDD-address-07` | When a second act map exists | normative | projected | +| P-3 | `DDD-address-08` | A process, and north | normative | projected | +| P-4 | `DDD-address-09` | The pull | conceptual | projected | +| P-5 | `DDD-address-10` | Kinds of demand as ground positions | conceptual | projected | +| P-6 | `DDD-address-11` | The four states | conceptual | projected | +| P-7 | `DDD-dec-122` | The decision that files them | decision | — | +| P-8 | — | The two statements of #80 that are stated, not ruled | — | — | + +## P-1 — `DDD-address-06`, the act map, amended + +**Weakest point.** The scope is declared before any act is mapped, as a set of decisions. The map's layers are composite acts, found only after. Nothing makes the scope's edge fall on composite acts. If it does not, a composite act straddles the edge, and its seam decisions are half inside and half world. + +**Statement.** The act map is the record of one ruled decomposition of the work inside one declared scope into acts, read under one declaration of ground axes, and it holds the act sites at their addresses, the composite acts that layer them, the causal dependencies between them and the index of the determinables that make demand arrive in its scope. + +**Region, in brief.** + +| Part | What it is | Rests on | +|---|---|---| +| Scope | The map's first declaration: the decisions its holders take. Everything outside is world to the map, including other people's decisions. A decision outside reaches the map only as ground crossing its edge | 11-18 R-2 | +| Act sites | One site per address: a ground position on the declared axes and a layer path. The occupant is left out. Occurrences are recorded beside the map, each against exactly one site, with its time index | `DDD-ground-06`; `DDD-address-01`; 11-17 R-7 point 1 | +| Layers | Composite acts above an act and below it, and no other kind of layer | 11-17 E-2; `DDD-ground-06` | +| Dependencies | B depends on A when a state change an occurrence at A produced is in the ground reading of an occurrence at B. A directed graph, not a tree | `DDD-address-01` | +| The index of determinables | One row per determinable at or inside the edge: its outcomes, its entry act, the act where its determinate lands, its specified, observed and expected states (actual is never read), its kind as a ground position, and its rate of change. A row whose specified state cites a decision in force is taken up | 11-17 R-3, R-6, R-7 point 2; 11-18 R-1, R-4, R-5 | + +The index's observed state is a pointer to the latest observation in the occurrence record, not a copy. The map stays structure. The record holds what changes. + +**Falsifier.** Any of: an act site that must sit within two composite acts for its governing set to be stated; a dependency that can be stated between two occurrences and between no two sites; a determinable that makes demand arrive in the scope and can be given no entry act on the map; or a composite act that cannot be placed wholly inside or wholly outside the declared scope. + +**Breaks.** The first three limbs break what they broke on 2026-11-17: the layer path, the split between map and record, and the claim that the map shows the pull. The fourth breaks scope as a declaration made before acts, and with it part 1 of the pull procedure. + +**Rests on.** `DDD-ground-06`, `DDD-ground-05`, `DDD-frame-13` (upstream, projected); `DDD-address-01` (projected). 11-17 E-2, R-3, R-6, R-7; 11-18 R-1, R-2, R-4, R-5. The fourth falsifier limb and the observed pointer rest on **no ruling**. + +**Notes it carries.** First drafted on 2026-11-17 on a branch that did not land, and amended on the rulings of 2026-11-18 (11-18 R-8). The 2026-11-17 text is on the archive branch at `47cf4cc`. + +## P-2 — `DDD-address-07`, when a second act map exists, amended + +**Weakest point.** **The two days' rulings conflict, and only Emil can settle it.** 11-17 R-7 point 3 rules that a second act map exists "when, and only when, a second declaration of ground axes exists". 11-18 R-2 rules scope the first declaration of a map, and R-3 speaks of a narrow scope's map and a wider map. The SDLC and the PDLC share one declaration of axes (`DDD-dec-115`). So under the 2026-11-17 wording, a map with the How as its scope and the PDLC's map are one map. Under R-3 they are two. + +**Two options.** + +| | Option A (recommended) | Option B | +|---|---|---| +| Statement | A second act map exists when, and only when, a second declaration exists: a second scope, or a second declaration of ground axes over one scope | As 2026-11-17: when, and only when, a second declaration of ground axes exists. A narrower scope under the same axes is a scope declared within the one map | +| R-3 reads as | Two maps: the narrow one and the wider one. The narrow scope's pull appears in the wider map as a dependency, and as a seam only where the wider map declares the narrow scope | One map, with nested scopes. "A wider map" means the map seen at its wider scope | +| What it supersedes | The wording of 11-17 R-7 point 3 | Nothing. But "everything outside the scope is world" then holds of a scope, not of a map | +| The lifecycle | Unchanged. `pdlc.md` §11's "one map, validated together" is one map with the PDLC as its scope, and the How declared within it | Unchanged | +| Weakest point of its own | It supersedes a ruled sentence | 11-18 R-2 calls scope "the first declaration of a map", which B demotes to a declaration within one | + +**Region, in brief, under A.** Two act sites belong to one map exactly when they are inside one declared scope and their positions are read on the same declared axes, under one ruled decomposition. A part of a map, such as the acts within one composite act, is not a map. A view is not a map. A wider map may declare a narrower scope within it. A pull of the narrower scope then shows as a seam at its edge. Where the wider map does not declare it, the pull shows as a dependency only (11-18 R-3). + +**Falsifier.** Unchanged in form: a governing decision whose applicability predicate must cover occurrences in two maps, or two occurrences in two maps whose behaviour must be compared to answer a question the work poses. Under A it adds: two maps of nested scopes whose shared acts carry different addresses. + +**Breaks.** Checking a map on its own (`applications/pdlc/graph/check.py`); the two grains of `DDD-address-05`; under A, any reading of the PDLC and the SDLC as one map that is then given two scopes. + +**Rests on.** 11-17 R-7 point 3; 11-18 R-2 and R-3; `DDD-ground-06`'s region; `DDD-address-04`; `DDD-dec-38` (the T2a map, three top acts under one declaration). That scope individuates a map rests on **no ruling** under either option until Emil rules Q-2. + +## P-3 — `DDD-address-08`, a process and north, amended + +**Weakest point.** "Unresolved" is now read as "two of the determinable's states disagree". A determinable can be unresolved with no gap anyone can read: actual differs from observed and nobody has looked. Such a north exists and no view can be fixed by it until someone observes. + +**Statement.** A process is a human understanding of work, scoped within an area of expertise and not an object of the model, and its translation into the model is a named view of one act map, fixed by one or more norths and optionally bounded by composite acts inside the map's scope, where a north is a determinable in the map's index that is unresolved: the determinable of a pull. + +**What changes from 2026-11-17.** Three things, each from a ruling of 2026-11-18. +- "Whether or not governance has taken it up" now means the governance inside the scope (#80, "State at opening"). A decision outside the scope is world, so a north may be taken up outside and still be a north of the map. +- A north is the determinable of a pull (P-4). The view runs from each north's entry act to the act where its determinate lands, both inside the scope. +- The bound lies inside the scope. A view cannot reach outside its map. + +Unchanged from 2026-11-17: a north is a cause, not a decision; the discharge act is the other end; time sets a determinable's rate of change, which is what creates demand; a view leaves out earlier readings; in tooling a process is a saved view, never a node. + +**Falsifier.** A view that the holders of a process do not recognise as their process; or a process whose holders can name no determinable it answers, once the calendar is set aside; or a process its holders recognise whose work answers no gap between any two states. + +**Breaks.** Step 1b's use of "a process" as its unit; the promotion condition of `DDD-flow-01` to `-04` ("a second process"); under the third limb, the identification of north with a pull's determinable. + +**Rests on.** 11-17 E-1, R-2, R-4, R-5, R-6, R-7 point 4; 11-18 R-2, R-5; `DDD-frame-13`; `DDD-address-01`; `DDD-ground-06`. That north is the determinable of a pull rests on **no ruling**: it is Claude's joining of 11-17 R-2 and 11-18 R-5, `UNVERIFIED — Emil review`. + +## P-4 — `DDD-address-09`, the pull, new + +**Weakest point.** The fourth gap, actual against observed, can never be inspected (`term:outcome`). So the most common deviation in software, the one nobody has seen, is a pull this claim names and its procedure cannot find. + +**Statement.** A pull is a gap between two of the four states of a determinable (`DDD-address-11`), read inside a map's declared scope, where the determinable has two or more outcomes that cannot all hold together, and a pull is placed by a trace back from the request that stops where the answer lies outside the scope. + +**Region, in brief.** Wherever a map's scope is declared. A pull is found at the edge first: what crosses into the scope and out of it (11-18 E-4, Claude's reading). Its entry act is the first act inside the scope that reads it. Under a different scope the same determinable may show a different gap and a different entry act (`pull-procedure.md`, W-1 and W-2). + +**Falsifier.** A request inside a declared scope that its holders recognise as demand and that shows no gap between any two states of any determinable at or inside the edge; or two practitioners who, given the same scope and the same record, independently write statements for one pull that name different entry acts. + +**Test.** Reproducibility, which is what Emil asked for at 11-18 E-3 ("so that it can be reproduced by practicitioners"): two practitioners run the five parts on one scope without conferring, and their statements are compared field by field. + +**Breaks.** Step 1 starting from scope and north (P-8, S-2); `DDD-address-08`'s north as a pull's determinable; the pull procedure. + +**Rests on.** 11-18 R-1, R-2, R-5; `DDD-frame-13` ("determination demand is thereby a measure over unresolved determinables"); `term:outcome`. The edge-first search rests on 11-18 E-4, which is not a ruling. + +## P-5 — `DDD-address-10`, kinds of demand as ground positions, new + +**Weakest point.** No map in the repository has declared the axis kinds are read on. The worked example declares two for exposition: realisation (not built, built) and whose determinable (the product's, the arrangement's). The second is needed only to place own-ground demand, and rests on no ruling. + +**Statement.** A kind of demand is a position on a declared ground axis, so kinds differ by the ground a demand pulls from and not by where it came from, and two demand items at one position are of one kind whatever their source. + +**Falsifier.** Two demand items at one ground position that must be ordered by different owners to be ordered on the ground they need; or one item whose kind changes with who reported it while its ground position is unchanged. + +**Breaks.** Step 1b's part 6 as written, which classifies each source of pull; the "sources by kind" tables of `pdlc.md` §9 and step 1b, read as classification by source. Not `DDD-flow-01`: its three kinds are already tested relative to the decisions in force, which is a ground reading. + +**Rests on.** 11-18 R-4; `DDD-ground-06` (positions on declared axes); consistent with `DDD-flow-01`. The second axis rests on **no ruling** (Q-4). + +## P-6 — `DDD-address-11`, the four states, new + +**Weakest point.** The mapping is built from four words (11-18, "The four states"). pjarnhus's own definitions are not known and may differ. "Expected" is not one state: each holder who bears the outcome has one. + +**Statement.** A determinable is read in four states inside a scope — specified, the governing decisions in force over it; actual, the outcome, the determinate as it lands in the world; observed, the verdict, the same determinate as assessed by a declared predicate; and expected, the tolerance held by each holder who bears the outcome — and actual is never read directly. + +**Credits** (the claim format's `credits` field, as `DDD-frame-13` uses it). The four state names, specified, actual, observed and expected, are GitHub user pjarnhus's, from unpublished work on four states of an implementation (11-18 R-6). The mapping of each to canon is ours, made from the names. + +**Falsifier.** A demand item whose gap cannot be stated without a fifth state; or a determinable outside software for which one of the four cannot be given any meaning. + +**Breaks.** `DDD-address-09`, which defines a pull over these states; the gap table; the index's state columns in `DDD-address-06`. + +**Rests on.** 11-18 R-5 and R-6; upstream `term:outcome`, `term:verdict`, `term:tolerance`, `term:governing-decision` (settled). "Never read directly" rests on `term:outcome` ("the world renders outcomes, never verdicts") and on an open point of #80. + +## P-7 — `DDD-dec-122`, amended + +One decision, Emil principal, recording the rulings of both days and the G1 rulings of this session in his words. It files the six claims, and the card changes Emil rules at G1. Proposed relations: + +| Relation | To | Why | +|---|---|---| +| Adds to | `DDD-dec-114` | Round 9's "the SDLC is a subprocess under the PDLC" keeps its ruled reading. In `DDD-address-08`'s terms the SDLC is a bound, and in `DDD-address-06`'s it may be declared a scope | +| Amends | `DDD-dec-118` | Its option text "Acts first, then how demand arrives" is amended if step 1 starts from scope and pull (S-2). Its "classify each source of pull" is amended by `DDD-address-10`. Its other rulings stand | +| Supersedes | Nothing by the `supersedes` field. Under P-2 option A, the wording of 11-17 R-7 point 3 | A decision's `supersedes` names a whole decision. Neither earlier decision is wholly replaced | + +`basedOn`: the six claims; `DDD-ground-05`, `DDD-ground-06`, `DDD-ground-01`, `DDD-frame-13` upstream; `DDD-address-01`, `-02`, `-04`; `DDD-flow-01`, `-03`; `DDD-dec-114`, `-118`, `-38`. + +Its notes say it was first drafted on 2026-11-17 on a branch that did not land and was amended on the rulings of 2026-11-18, and they supersede 11-17 R-7 point 7 ("no issue number on the commits"), as the 2026-11-18 record's first correction already says. + +## P-8 — the two statements that are stated, not ruled + +#80 lists them as stated and not ruled. They are proposed here and are not ground. + +| # | Statement in #80 | Proposed reading | Falsifier | Rests on | +|---|---|---|---|---| +| S-1 | "The act map is the central modelling artefact for every application" | Central as an index every later artefact attaches to, not as a container that holds them. `steps-and-the-map.md` is the evidence: each of steps 2 to 10 attaches to a part of the map or to a record beside it | An artefact a later step produces that has no part of the map, and no record beside it, to attach to | 11-18 E-1, not ruled | +| S-2 | "Step 1 starts from north" | Step 1 starts from scope, then north. Scope comes first because 11-18 R-2 says "scope must be the first else we cant place the pull". North is found by the five-part procedure, before any act is mapped | A pull statement whose entry act cannot be found until acts are mapped | 11-18 E-2, not ruled; `DDD-ground-05` (declaring the determinable space precedes determination over it) | + +**Where the procedure is filed (S-2).** + +| | Option A (recommended) | Option B | +|---|---|---| +| Form | The five parts become the first part of step 1's card, by a superseding section, marked `[PROPOSED]` inside a card ruled for test | A card of its own before step 1, `[PROPOSED]`, numbered when N-1 is ruled | +| For it | It is what Emil stated at 11-18 E-2, and it keeps one step for one map | Step 1's ruled form is untouched, and its register is not mixed | +| Against it | One card carries two registers | Step 1b's numbering problem doubles | +| T2b | Its act map is produced by step 1 with the new first part. See `effects.md` | T2b runs step 1 unchanged. The pull procedure is tested separately | + +## G1 questions + +| # | Question | +|---|---| +| Q-1 | P-1: is `DDD-address-06` ruled as amended, with the index carrying the three readable states and the observed state as a pointer into the record? | +| Q-2 | P-2: option A or option B for `DDD-address-07`? Under A, the wording of 11-17 R-7 point 3 is superseded | +| Q-3 | P-3: is `DDD-address-08` ruled as amended, with north as the determinable of a pull? | +| Q-4 | P-5: does each map declare the axis its kinds are read on, or does the method give one? Is "whose determinable" ruled as the place of own-ground demand (`pull-procedure.md`, W-3)? | +| Q-5 | P-4, P-5, P-6: are `DDD-address-09` to `-11` ruled as written, each conceptual and projected? | +| Q-6 | Does 11-18 R-5 rule the gap table as well as the mapping of the four states? If not, is it ruled now, and is "specified against observed" ruled as the readable form of "specified against actual"? | +| Q-7 | May a file under `core/claims/` carry pjarnhus's credit, given that his agreement is not recorded (an open point of #80)? If not, `DDD-address-11` carries "the four state names are not ours" with no name, and the credit stays in the session record | +| Q-8 | P-8: S-1 and S-2 as read? For S-2, option A or option B? | +| Q-9 | `steps-and-the-map.md`: are the nine "Value made visible" rows ruled as proposed? | +| Q-10 | `effects.md`: is the effect on T2b, on step 1b's kinds and on `DDD-address-02`'s region ruled as stated, and which of the three options for `DDD-address-02`? | +| Q-11 | Filing: the six claims under `core/claims/`, `DDD-dec-122` under `core/decisions/`, the card changes in `method/`, and this session's row in `meta/sessions/README.md`. Anything else, or anything not? | + +## Amendments after G1 + +Appended on Emil's G1 rulings of 10:30 (`rulings.md`). The sections above are left as written, and where they differ, this section governs. These are the texts that will be filed once the held conflicts K-1 to K-3 are answered. + +### G1-A1 — P-2, `DDD-address-07`, on C-1 and G1-Q2 + +**Statement, ruled (option A).** A second act map exists when, and only when, a second declaration exists: a second scope, or a second declaration of ground axes. Scope says what is world. Axes say what is comparable. + +**Falsifier, corrected by C-1.** Two limbs, one per kind of boundary. +- Between maps with different declarations of ground axes: a governing decision whose applicability predicate must cover occurrences in both, or two occurrences in the two maps whose behaviour must be compared to answer a question the work poses. +- Between maps of nested scopes on the same axes: shared acts that carry different addresses in the two maps. + +The first limb no longer fires on a PDLC decision covering an SDLC act, which is the case C-1 names. + +### G1-A2 — P-3, `DDD-address-08`, on C-2. **Held: K-1** + +The text C-2 asks for, to be filed once K-1 is answered: "… fixed by one or more norths and optionally bounded by composite acts inside the map's scope, where a north is the determinable of a pull, whether or not a gap is open on it now." A process keeps its north between occurrences: the determinable is to the gap as the act site is to the occurrence. The falsifier's third limb, "a process its holders recognise whose work answers no gap between any two states", reads "no gap at any time", not "no gap now". + +### G1-A3 — P-4 and P-6, on G1-Q6 and G1-W + +`DDD-address-09`'s region says: a pull is a gap between two of the four states; the gap table is examples; "specified against observed" is the readable stand-in for "specified against actual"; "found" means any holder observing an outcome outside their tolerance, which is "expected against observed", found from the expected side. Its weakest point keeps the unreadable gap between actual and observed. Whether it adds "a bug never found has no impact" is held at K-3. + +`DDD-address-11`'s credits name pjarnhus, without an @, and its notes record that Emil told him and he agreed (G1-Q7). + +### G1-A4 — P-5, on G1-Q4 + +`DDD-address-10`'s region says: each map declares the axis its kinds are read on, and the method offers "built or not built" as the starting point. "Whose determinable" is named in its notes as `[PROPOSED]`, with own-ground demand as its only evidence. diff --git a/meta/sessions/2026-11-18-act-map-scope-and-pull/pull-procedure.md b/meta/sessions/2026-11-18-act-map-scope-and-pull/pull-procedure.md new file mode 100644 index 0000000..d65074a --- /dev/null +++ b/meta/sessions/2026-11-18-act-map-scope-and-pull/pull-procedure.md @@ -0,0 +1,190 @@ +# The pull statement and the five-step procedure + +A pull is a gap between two of four states of a determinable, read inside a declared scope. The four states are specified, actual, observed and expected. **The four states are credited to GitHub user pjarnhus** (2026-11-18 R-6). Only their four names are his. The mapping to canon, and everything below, is ours. + +The procedure finds the pulls of a scope before any act inside it is mapped. It has five parts, in this order: scope, the determinables at the edge, the three readable states, the gaps, and the trace to an entry act. Its output is one pull statement per pull. Step 1 then maps acts from the entry acts the statements name. + +**Register: `[PROPOSED]`.** Nothing here is filed. Each part names the ruling it rests on. Where a part rests on no ruling, it says so. + +**Weakest point.** "Actual" is never read directly (`term:outcome`: "the world renders outcomes, never verdicts"). So the gap between specified and actual, which is the bug, is only ever read through observed. A practitioner who has not observed a deviation cannot find it with this procedure. The procedure finds pulls someone has seen or expects. It cannot find the ones nobody has seen. + +## The five parts + +| # | Part | The question, in plain words | Rests on | +|---|---|---|---| +| 1 | **Scope** | Whose decisions are we mapping? Everything else is the world | 2026-11-18 R-2: scope is the first declaration of a map. Everything outside it is world, including other people's decisions | +| 2 | **The determinables at the edge** | What comes into the work and what goes out? For each, what could turn out more than one way? | 2026-11-18 R-1: two or more outcomes that cannot all hold together. Looking at the edge first is E-4 ("they also inspect a system and look at the input and output"), read by Claude, `UNVERIFIED — Emil review` | +| 3 | **The three readable states** | What was decided? What has someone seen? What does each person who bears it expect? | 2026-11-18 R-5, the mapping to canon. That "actual" is not readable is an open point of #80 | +| 4 | **The gaps** | Which two states disagree? | 2026-11-18 R-5: a pull is a gap between two of the four states. The table of gaps is Claude's (see below) | +| 5 | **The trace to an entry act** | Follow the gap back. Stop where someone outside the scope would have to answer. Which act inside the scope takes it in? | 2026-11-18 R-2: the trace back from a request stops where the answer lies outside the scope | + +## The pull statement + +One statement per pull. Every field has a check that passes or fails. "Mechanical" means a reader can check it against the statement alone. "Judgement" means the check needs someone who knows the work. + +| # | Field | What it holds | Check: it passes when | Kind of check | Part | +|---|---|---|---|---|---| +| F-1 | Scope | The name of the scope, and the decisions inside it: those the map's holders take | Every decision the statement names is marked inside or outside, and none is marked both | Mechanical | 1 | +| F-2 | Edge crossing | What crosses the edge, in or out, with the side that produces it and the side that reads it | Exactly one side is inside the scope | Mechanical | 1, 2 | +| F-3 | Determinable | The dimension that can turn out more than one way, named in words | It is about something that crosses the edge in F-2 | Mechanical | 2 | +| F-4 | Outcomes | Two or more outcomes of F-3 | There are at least two, and no two can hold together at the grain the statement names | Judgement | 2 | +| F-5 | Specified | The governing decision in force that covers F-3, cited, with its holder marked inside or outside; or "none" | A citation that resolves to a decision, or the word "none" | Mechanical | 3 | +| F-6 | Observed | What someone inside the scope has seen, by whom and in which record; or "not observed" | The observer is inside the scope. A sighting made outside is written as an edge crossing in F-2, not here | Mechanical | 3 | +| F-7 | Expected | One row per holder: who bears the outcome, and the outcome they expect | Each row names its holder. Two holders who disagree are both kept | Mechanical | 3 | +| F-8 | Actual | Always "not read" | The field says "not read". Anything else fails: a reading is F-6 | Mechanical | 3 | +| F-9 | Gap | The two states that disagree, as a row of the gap table | Both states are filled, or the row allows "none". A gap with actual in it is marked "suspected" | Mechanical | 4 | +| F-10 | Kind | The position on a declared ground axis that the pull pulls from, naming the axis | The axis is named and the position is one of its values. Where the item came from is not used | Mechanical once the axis is declared. Declaring the axis is judgement | 4 | +| F-11 | Rate of change | How often time makes F-3 unresolved again, with a unit, or "no unit known" | It is stated, with a unit or the words "no unit known" | Mechanical | 4 | +| F-12 | Trace and stop | The chain followed back from the gap, and the point where it stops because the answer lies outside the scope | Every link but the last is inside the scope. The last is outside and is named as world, or the chain ends inside at the cause | Judgement | 5 | +| F-13 | Entry act | The first act inside the scope that reads the pull | It is inside the scope, and the trace in F-12 reaches it | Mechanical | 5 | + +F-11 is the rate of change of 2026-11-17 R-6, carried here because the index holds it. Whether a determinable is taken up by governance is read from F-5: a decision in force means taken up. + +## The gap table + +`[PROPOSED]`. Claude's table from the 2026-11-18 record, carried unchanged. R-5 rules the mapping of the four states. Whether it also rules this table is a question for G1 (Q-6 in `proposals.md`). + +| Gap | What it is | Readable? | +|---|---|---| +| Expected, with nothing specified | A need nobody has taken up: a feature | Yes | +| Specified, with nothing actual | An approved change not yet built | Only as specified with nothing observed | +| Specified against actual | A deviation: a bug, which needs an implementation to exist | Only as specified against observed | +| Actual against observed | A deviation nobody has seen, or an alarm with nothing behind it | No. It can be suspected, never inspected | +| Expected against specified | The specification is wrong | Yes | + +Two pairs are not in the table: observed against expected, and specified against observed read as itself. The worked example uses the second as the readable form of "specified against actual". That reading is Claude's, `UNVERIFIED — Emil review`. + +## Worked example: the acme shop, with the How as the scope + +The acme shop is the invented, illustrative record of [`applications/pdlc/`](../../../applications/pdlc/). The scope is the one Emil gave as his example in 2026-11-18 R-2: the How, "which is essentially the SDLC". The example is carried through all five parts. One item, issue #103, is followed in full. Two more, #101 and #104, are carried beside it, because kinds can only be told apart by comparison. + +### Part 1 — scope + +| Field | Value | +|---|---| +| F-1 Scope | **The How.** Inside: the decisions the How owner holds, about how the shop's software is planned, built, accepted, landed and run (pdlc.md §2, §13) | +| Outside, as world | The When owner's decisions: what is released for work, in which order, and when a toggle goes on. The What owner's decisions: what the software must do, its acceptance, and whether a defect is a bug. Customers' decisions | + +| F-2 Edge crossing | Produced by | Read by | Source | +|---|---|---|---| +| In: a released issue (new demand) | When owner, outside | How owner, inside | pdlc.md §13 | +| In: an owed issue, with the classification verdict | What owner, outside | How owner, inside | pdlc.md §13 | +| In: the What and its product assurance | What owner, outside | How owner, inside | pdlc.md §13 | +| Out: production state, dark | How owner, inside | What owner, outside | pdlc.md §13 | +| Out: telemetry of the product in use | How owner's instrumentation, inside | What owner, outside. The How owner reads it too | pdlc.md §13; `DDD-flow-03` (sighting) | +| Out: the record, with the DORA keys | How owner, inside | When owner, outside | pdlc.md §13 | + +Check F-1 passes: every decision named is marked once. Check F-2 passes: each row has one side inside. + +### Part 2 — the determinables at the edge + +| Item | F-3 Determinable | Crosses at | F-4 Outcomes | Check F-4 | +|---|---|---|---|---| +| #103 | Whether reorder respects the stock invariant | In: the What (the invariant); out: telemetry (the basket) | (a) reorder never adds an out-of-stock line; (b) reorder adds an out-of-stock line | Passes. (a) and (b) cannot both hold for one basket | +| #101 | Whether a returning customer can reorder in one step | In: a released issue | (a) in one step; (b) only by building the basket again | Passes | +| #104 | Whether a deploy needs a manual cache purge | Out: the record (each deploy) | (a) no manual step; (b) a manual purge on every deploy | Passes | + +A rejected candidate shows the check failing. "The shop's quality" names no outcomes that exclude each other, so it is not a determinable and gets no statement. + +### Part 3 — the three readable states + +| Item | F-5 Specified | F-6 Observed | F-7 Expected | F-8 Actual | +|---|---|---|---|---| +| #103 | The Place order Decider's stock invariant (example-pool.ttl, the report on #103). Holder: What owner, **outside** | The out-of-stock line in a basket, on the How owner's telemetry (demand-flow.ttl, `t103`) | Customers: (a). What owner: (a) | Not read | +| #101 | The What decision framed at o1 for the issue. Holder: What owner, **outside** | Not observed: nothing built yet | Customers, through the When owner's market need: (a) | Not read | +| #104 | None found in the record | A manual purge on every deploy, in the How owner's own record | How owner: (a) | Not read | + +Checks pass for all three rows. #104's "none" is a finding about the record, not a failure: nobody decided that deploys need a purge. It is how the arrangement came to be. + +### Part 4 — the gaps + +The kind is read on one declared axis, which the worked example declares here. It is Claude's, for exposition, built from Emil's ruling that "ground for a feature is by necessity that it hasnt been implemented and a bug must have been implemented" (R-4). `UNVERIFIED — Emil review`. + +| Axis | Values | Why it is an axis | +|---|---|---| +| Realisation of the determinable in the scope | not built · built | R-4: a feature and a bug cannot share ground, because one has not been implemented and the other has | +| Whose determinable | the product's · the arrangement's | Without it #104 has no place (see the finding below) | + +| Item | F-9 Gap | Readable as | F-10 Kind: position | F-11 Rate of change | +|---|---|---|---|---| +| #103 | Specified against actual: a bug | Specified against observed | product · built | Per reorder of a basket holding an item since sold out. Unit: reorders | +| #101 | Specified, with nothing actual: an approved change not yet built | Specified, with nothing observed | product · not built | Set by the market. No unit known | +| #104 | Expected, with nothing specified | As itself | arrangement · built | Per deploy. Unit: deploys | + +Read in the lifecycle's words, #103 is owed, #101 is new and #104 is own-ground. The positions tell them apart without asking who reported them. That is R-4. + +### Part 5 — the trace to an entry act + +| Item | F-12 Trace back from the request, and where it stops | F-13 Entry act | +|---|---|---| +| #103 | The request to fix reorder ← the owed verdict: is the out-of-stock line a bug, or is the What wrong? That answer is the What owner's. **Stop: outside the scope.** The verdict arrives across the edge as ground | Plan the realisation (o1b in the acme map), which reads the owed verdict | +| #101 | The released issue ← the When owner's order and the What owner's framing. **Stop: outside** | Plan the realisation (o1b), which reads the framing | +| #104 | The request to remove the purge ← the How owner's own record of deploys. The record is inside. The chain ends inside, at the cause | Frame own-ground work (oG) | + +**The same item under a wider scope.** With the PDLC as the scope, the What owner's classification is inside. #103's trace then runs on past it, to the product in use and the customer's basket, and stops at the customer, who is world. The entry act becomes oC, the classification, which is what `DDD-flow-03` names. The What → How seam is visible in that map only because the PDLC map declares the How as a scope within it. That is R-3: "if we started with the PDLC we wouldnt see the seam in the how because it wouldnt exist" read the other way round. + +## Desk check against a junior engineer's first contact + +Claude's desk check, `UNVERIFIED — Emil review`. It is a reading of this page, not a run. Running the junior-engineer test is Emil's act, as it is for the quick start. + +| Part | What the junior engineer must do | Passes at first contact? | What fails, and the fix proposed | +|---|---|---|---| +| 1 | List the decisions their own team makes | **Fails as written.** "The How owner", "the What owner" and "the When owner" are lifecycle words | Ask "which decisions can your team make without asking anyone?" and list them. The owner names follow | +| 2 | Name what crosses into and out of the team's work, and two outcomes for each | Passes. "Name two ways it could turn out that cannot both be true" needs no canon word | Keep the word "determinable" out of the question and in the field name | +| 3 | Find the decision in force; say what they have seen; ask who bears it | Partly. Specified needs to know where decisions are written down. A junior engineer in a team without a written What fails here | Allow "none written" as a value. That is a finding, as #104's is | +| 3 | Write "not read" under actual | Passes mechanically, and confuses. Why would a field always say the same thing? | One sentence: "you only ever see what someone observed; the world itself is not a reading" | +| 4 | Pick the row of the gap table | Passes mechanically | — | +| 4 | Read the kind off the axis | **Fails without the axis.** The junior engineer cannot invent "realisation" | The procedure must give the axis, or step 1 must declare it first. This is Q-4 in `proposals.md` | +| 5 | Follow the chain back and stop at the first answer outside the team | Passes when part 1 passed. Fails with it | Same fix as part 1 | + +## Findings + +| # | Finding | Status | +|---|---|---| +| W-1 | The entry act depends on the scope. Under the How, #103 enters at plan the realisation. Under the PDLC it enters at the classification. `DDD-flow-03` is scoped to the PDLC and is not contradicted | Claude's reading, `UNVERIFIED — Emil review` | +| W-2 | The same determinable shows a different gap under a different scope. Under the PDLC, #101 is first "expected, with nothing specified" at the option pool. Under the How it is "specified, with nothing actual". Its position, product · not built, is the same under both | Claude's reading, `UNVERIFIED — Emil review` | +| W-3 | Own-ground demand finds a place among the gaps once the determinable is the arrangement's, not the product's. #104 is "expected, with nothing specified" on the arrangement. A dependency's CVE would be "specified against actual" on the arrangement. So own-ground differs from new and owed by whose determinable it is, not by its gap. This answers an open point of #80 with a proposal, not a ruling. #104 also shows a combination the PDLC's kinds do not name: built, with nothing specified. The arrangement behaves a way nobody decided | `[PROPOSED]`, `UNVERIFIED — Emil review` | +| W-4 | The kind needs a declared axis before part 4, and the procedure runs before step 1 declares axes. So the procedure declares at least the axis kinds are read on | `[OPEN]`, Q-4 | +| W-5 | The rate of change has a unit for two items and none for the market need | Carried from `successor-items.md` O-7 of 2026-11-17: no unit and no measure | + +## Amendments after G1 + +Appended on Emil's G1 rulings of 10:30 (`rulings.md`). The sections above are left as written, and where they differ, this section governs. + +### G1-A1 — the desk check was a reading, not a run + +The desk check above is Claude's reading of this page. No junior engineer has run the procedure. Running the junior-engineer test is Emil's act, and it has not been done (G1-D). + +### G1-A2 — part 1 in plain words + +Part 1's question, as it goes on the card, names no owner (G1-D): + +> **Which decisions can the people doing this work make without asking anyone else?** List them. That list is the scope. Every other decision, and everything nobody decides, is the world. The world still reaches you: through what comes in across the edge. + +Worked example, re-run with that question. The team that builds and runs the shop's software can decide, without asking anyone, how a change is planned, built, checked, landed and deployed. It cannot decide what the shop must do, which work comes first, or when a feature is switched on for customers. So F-1 is: inside, planning, building, checking, landing and deploying changes; outside, what the shop must do, the order of work, switching features on, and customers' choices. The edge crossings of F-2 are unchanged and still have one side inside each. Checks F-1 and F-2 pass. + +### G1-A3 — part 4's axis + +Each map declares the axis its kinds are read on (G1-Q4). The method offers **built or not built** as the starting point, as it offers the PDLC's kinds. "Whose determinable" stays `[PROPOSED]`, with own-ground demand as its only evidence. + +| Item | Position on built or not built | Kind told apart? | +|---|---|---| +| #101 | not built | Yes, from #103 | +| #103 | built | Yes, from #101 | +| #104 | built | **No.** On this axis alone it sits with #103. It is told apart only by the `[PROPOSED]` axis "whose determinable", or by a map that declares its own | + +### G1-A4 — the gap table, as ruled + +The definition is ruled: a pull is a gap between two of the four states (G1-Q6). The table is examples, not an exhaustive list. One row is added: + +| Gap | What it is | Readable? | +|---|---|---| +| Expected against observed | The software matches its specification, and the holder still cannot do the thing | Yes | + +"Specified against observed" is the readable stand-in for "specified against actual". The claim calls it a stand-in. + +**"Found", as ruled (G1-W).** Any holder observing an outcome outside their tolerance has found it, whether or not they trace it to the software. That is "expected against observed", and the procedure finds it from the expected side. In the worked example, #103 is also found this way: a customer sees the out-of-stock line in the basket. #104 is too: the people who deploy observe a manual purge they do not expect. + +### G1-A5 — the credit + +Emil has told pjarnhus, and he agreed to the four states appearing here under his credit (G1-Q7). The open point of #80 that his agreement was not recorded is closed. diff --git a/meta/sessions/2026-11-18-act-map-scope-and-pull/rulings.md b/meta/sessions/2026-11-18-act-map-scope-and-pull/rulings.md new file mode 100644 index 0000000..24ef96b --- /dev/null +++ b/meta/sessions/2026-11-18-act-map-scope-and-pull/rulings.md @@ -0,0 +1,227 @@ +# Rulings — the act map, scope and pull, 2026-11-18 + +Emil's G1 rulings, received in this session at 10:30 Europe/Berlin (09:30 UTC) on 2026-11-18. His message is verbatim below. The reading table after it is Claude's, and each reading Emil has not confirmed is marked. + +## Emil's message, verbatim + +> My G1 rulings on #80. Record this message in the session's rulings.md, in my words, with the +> time you receive it. Then do deliverable 6 and stop at G2. +> +> Two corrections first. Make them in the proposals before filing. +> +> C-1, on P-2. The falsifier limb "a governing decision whose applicability predicate must cover +> occurrences in two maps" applies only to maps with different declarations of ground axes. For +> nested scopes on the same axes the falsifier is the new limb: shared acts that carry different +> addresses. As written, any PDLC decision covering an SDLC act fires it on day one. +> +> C-2, on P-3. A north is the determinable of a pull whether or not a gap is open right now. +> Drop "that is unresolved" as a condition. A process keeps its north between occurrences. The +> determinable is to the gap as the act site is to the occurrence. +> +> The rulings. +> +> Q-1. Yes. DDD-address-06 as amended, with the three readable states in the index, the observed +> state as a pointer into the occurrence record, and the fourth falsifier limb. +> +> Q-2. Option A, with C-1. A second act map exists when a second scope or a second declaration +> of ground axes exists. Scope says what is world. Axes say what is comparable. This supersedes +> the wording of 2026-11-17 R-7 point 3. +> +> Q-3. Yes, with C-2. North is the determinable of a pull. +> +> Q-4. Each map declares the axis its kinds are read on. The method offers "built or not built" +> as the starting point, as the PDLC's kinds are a starting point. "Whose determinable" stays +> [PROPOSED]: own-ground demand is its only evidence. +> +> Q-5. Yes. DDD-address-09, -10 and -11 as written, each conceptual and projected. +> +> Q-6. I rule the definition: a pull is a gap between two of the four states. The gap table is +> examples, not an exhaustive list. Add "expected against observed" to it: the software matches +> its specification and the holder still cannot do the thing. "Specified against observed" is +> the readable stand-in for "specified against actual", and the claim calls it a stand-in. +> +> Q-7. Yes. DDD-address-11 carries pjarnhus's credit by name, written without an @. I have told +> him and he agreed. Record that, and close the open point of #80 that said his agreement was +> not recorded. +> +> Q-8. S-1 as read: the act map is central as an index every later artefact attaches to, not as +> a container. S-2 as read, option A: the five parts become the first part of step 1's card, by +> a superseding section, marked [PROPOSED] inside the card. +> +> Q-9. Yes, as [PROPOSED], with one change. Row 5 attaches closure to the act sites, not to the +> index's observed state. Closure is a property of an act's tolerance. +> +> Q-10. Yes. The effects on T2b and on step 1b's kinds as stated. Option 1 for DDD-address-02: +> a note appended to the claim, the region not reworded. +> +> Q-11. As listed: the six claims under core/claims/, DDD-dec-122 under core/decisions/, the card +> changes in method/, and the session's row in meta/sessions/README.md. Nothing else. +> +> On the weakest point of the pull, my words: "a bug nobody has found will never be found. and a +> bug that is never found has no impact." So the unreadable gap between actual and observed is +> not a defect of the procedure. Say this on step 1's card, not only in the claim. +> I also rule the reading: "found" means any holder observing an outcome outside their +> tolerance, whether or not they trace it to the software. Such a gap is "expected against +> observed", and the procedure finds it from the expected side. +> +> On the desk check. Part 1 of the procedure must state scope in plain words, without the owner +> names. Part 4 gets its axis from Q-4. Fix both before the card is filed, and say in the session +> record that the desk check was a reading and not a run. +> +> Filing. +> - Ids as ruled on 2026-11-18 R-8: -06, -07, -08 and DDD-dec-122 amended before landing, with +> the note on their history. New claims -09, -10, -11. +> - Change ruled material by supersession, never by rewriting. +> - Every commit carries "Refs #80". Commits that change core/ carry "Basis:" and name the ids +> they touch. +> - Run the whole validate workflow locally before each push, and report the workflow run for +> the pushed head. +> - Do not touch the archive branch, tools/canon, any act or any lifecycle data. +> +> Stop at G2. Open a pull request against main and name the head to ratify. Merging is mine. +> If anything here conflicts with a filed claim or an earlier ruling, stop and tell me before +> filing. + +## Reading + +| # | Time | Status | Reading | +|---|---|---|---| +| C-1 | 10:30 | **Correction by Emil**, on P-2 | The limb "a governing decision whose applicability predicate must cover occurrences in two maps" holds only between maps with different declarations of ground axes. Between nested scopes on the same axes, the falsifier is shared acts carrying different addresses | +| C-2 | 10:30 | **Correction by Emil**, on P-3. **Held: see "Conflicts found before filing", K-1** | A north is the determinable of a pull whether or not a gap is open now. "That is unresolved" is dropped as a condition. Determinable is to gap as act site is to occurrence | +| G1-Q1 | 10:30 | **Ruled** | `DDD-address-06` as amended (P-1), with the three readable states in the index, the observed state as a pointer into the occurrence record, and the fourth falsifier limb | +| G1-Q2 | 10:30 | **Ruled**, supersedes the wording of 11-17 R-7 point 3 | Option A with C-1: a second act map exists when a second scope or a second declaration of ground axes exists. "Scope says what is world. Axes say what is comparable" | +| G1-Q3 | 10:30 | **Ruled**, with C-2. **Held with C-2** | North is the determinable of a pull | +| G1-Q4 | 10:30 | **Ruled** | Each map declares the axis its kinds are read on. The method offers "built or not built" as the starting point. "Whose determinable" stays `[PROPOSED]` | +| G1-Q5 | 10:30 | **Ruled** | `DDD-address-09`, `-10`, `-11` as written, each conceptual and projected | +| G1-Q6 | 10:30 | **Ruled** | The definition: a pull is a gap between two of the four states. The gap table is examples, not exhaustive. "Expected against observed" is added. "Specified against observed" is the readable stand-in for "specified against actual", and the claim calls it a stand-in | +| G1-Q7 | 10:30 | **Ruled** | `DDD-address-11` credits pjarnhus by name, without an @. **Emil has told pjarnhus and he agreed.** The open point of #80, that his agreement was not recorded, is closed | +| G1-Q8 | 10:30 | **Ruled** | S-1 as read. S-2 as read, option A: the five parts become the first part of step 1's card, by a superseding section marked `[PROPOSED]` | +| G1-Q9 | 10:30 | **Ruled**, `[PROPOSED]` rows. **Its reason held: K-2** | The nine rows as proposed, except that row 5 attaches closure to the act sites, not to the index's observed state | +| G1-Q10 | 10:30 | **Ruled** | The effects on T2b and on step 1b's kinds as stated. Option 1 for `DDD-address-02` | +| G1-Q11 | 10:30 | **Ruled** | Filing as listed, and nothing else | +| G1-W | 10:30 | **Ruled**, the reading of "found". **Its first sentence held: K-3** | "Found" means any holder observing an outcome outside their tolerance, whether or not they trace it to the software. Such a gap is "expected against observed", found from the expected side. The unreadable gap between actual and observed is not a defect of the procedure, and step 1's card says so | +| G1-D | 10:30 | **Ruled** | Part 1 states scope in plain words, without the owner names. Part 4 takes its axis from Q-4. The session record says the desk check was a reading, not a run | + +## Conflicts found before filing + +Emil's instruction: "If anything here conflicts with a filed claim or an earlier ruling, stop and tell me before filing." Three were found. Nothing under `core/` or `method/` is filed until he answers. Canon is read at `v5.15.2`. + +| # | Ruling | Conflicts with | The conflict | +|---|---|---|---| +| K-1 | C-2 and G1-Q3: drop "that is unresolved" | 11-17 R-2, ruled: "North is the unresolved determinable". #80, "Ruled before this issue was opened": "North is an unresolved determinable" | The earlier ruling makes "unresolved" part of what north is. C-2 removes it. G1-Q2 names the earlier wording it supersedes; C-2 does not. C-2 agrees with 11-17 R-6 (time makes a determinable unresolved again at its rate), so this may be a supersession of R-2's wording and not a contradiction of its sense. Claude's reading, `UNVERIFIED — Emil review` | +| K-2 | G1-Q9's reason: "Closure is a property of an act's tolerance" | Upstream `term:closure`, settled: "A predicate is **closed for an arrangement** when the relevant ground is observable and adequacy can be evaluated within declared resource, latency, and confidence bounds" (`core/03-the-floor.md`, "The claim", "Effective closure, defined") | Canon makes closure relative to the arrangement, not to the tolerance alone. `DDD-ground-06` keeps the occupant out of the address, so an act site alone does not fix closure. And step 5 comes before step 6 (allocate), where the arrangement is placed | +| K-3 | G1-W: "a bug that is never found has no impact", on step 1's card | Upstream `term:escape`, settled: "determined *never*, by nobody … Latent defect exposure. **The only forbidden state.**" (`core/01`). And 11-17 E-4, Emil's own: "a fire in a forest creates demand even though no one is there" | Canon names latent, unfound exposure as the forbidden state. Writing on a card that an unfound bug "has no impact" reads against it. The ruled reading of "found" narrows the gap but does not close it: an exposure no holder has yet observed is latent until one does | + +## Corrections + +- **Emil, on P-2 (C-1).** Claude's falsifier for `DDD-address-07` kept the 2026-11-17 limb for every pair of maps. Under option A it fires on any PDLC decision covering an SDLC act. Corrected: the limb holds only between different declarations of axes. +- **Emil, on P-3 (C-2).** Claude made "unresolved" a condition of north. Emil drops it. Held at K-1. +- **Emil, on the desk check.** Claude's `pull-procedure.md` recorded the desk check as a reading. Emil's ruling makes the record say so plainly: the junior-engineer desk check was a reading of the page by Claude, not a run with a junior engineer. Nobody has run the test. +- **Emil, on Q-9 row 5.** Claude attached closure to the index's observed state. Emil moves it to the act sites. + +## Emil's answers to K-1 to K-3, verbatim + +Received at 13:26 Europe/Berlin (12:26 UTC) on 2026-11-18. + +> My answers to K-1, K-2 and K-3. Record this message in rulings.md, in my words, with the time +> you receive it. Then file. +> +> K-1. Yes. C-2 supersedes the wording of 2026-11-17 R-2, not its meaning. A north is a cause and +> a determinable, as R-2 ruled. "Unresolved" is dropped as a condition: a north is the +> determinable of a pull whether or not a gap is open right now. Record it as a supersession of +> R-2's wording, and of the matching line in #80's "ruled before" list. It rests on 2026-11-17 +> R-6. +> +> K-2. Take your option. Row 5 attaches closure to the act site's acceptance predicate, closed +> for the arrangement allocated there (term:closure). I withdraw the reason I gave under Q-9, +> "closure is a property of an act's tolerance": it left the arrangement out. This is not a canon +> question. Note as an open point, without resolving it here, that step 5 classifies closure +> before step 6 allocates the arrangement. +> +> K-3. Step 1's card does not say that an unfound deviation has no impact. It says three things: +> - the pull procedure finds a deviation once a holder observes an outcome outside their +> tolerance, from the expected side; +> - exposure that nobody has observed is latent, which canon calls escaped (term:escape), and +> this procedure does not find it; +> - finding it before it lands is the work of assurance, in step 5, not of finding pulls. +> My sentence "a bug nobody has found will never be found. and a bug that is never found has no +> impact" stays in the record as I said it. It holds of a deviation that is never found by +> anyone. Nobody can know "never" in advance, so it is not a rule for the card. My ruled reading +> of "found" stands. The weakest point of DDD-address-09 stays as you wrote it. +> +> With these three answered, file as ruled at G1: the six claims, DDD-dec-122, the card changes +> and the README row. Record each of these as a correction with its attribution: K-2's reason +> and K-3's sentence are mine, and you caught both. +> +> Stop at G2. Open the pull request against main and name the head to ratify. + +| # | Time | Status | Reading | +|---|---|---|---| +| K-1 | 13:26 | **Ruled.** Supersedes the wording of 11-17 R-2 and the matching line of #80's "Ruled before this issue was opened" list ("North is an unresolved determinable, where demand originates"), not their meaning | A north is a cause and a determinable. It is the determinable of a pull whether or not a gap is open now. Rests on 11-17 R-6. C-2 and G1-Q3 are released from hold | +| K-2 | 13:26 | **Ruled.** Emil's Q-9 reason withdrawn | Row 5 attaches closure to the act site's acceptance predicate, closed for the arrangement allocated there (`term:closure`). Open point, not resolved here: step 5 classifies closure before step 6 allocates the arrangement | +| K-3 | 13:26 | **Ruled** | Step 1's card says three things: the procedure finds a deviation once a holder observes an outcome outside their tolerance, from the expected side; unobserved exposure is latent, which canon calls escaped (`term:escape`), and the procedure does not find it; finding it before it lands is assurance's work, in step 5. The card does not say an unfound deviation has no impact. `DDD-address-09`'s weakest point stands as written | +| F | 13:26 | **Ruled** | File as ruled at G1: the six claims, `DDD-dec-122`, the card changes, the README row | + +## Corrections, after K-1 to K-3 + +- **Claude, on Emil's Q-9 reason (K-2).** Emil gave "closure is a property of an act's tolerance". Claude caught that upstream `term:closure` makes a predicate "closed for an arrangement". Emil withdrew the reason: "it left the arrangement out". +- **Claude, on Emil's sentence for step 1's card (K-3).** Emil asked the card to say "a bug that is never found has no impact". Claude caught that upstream `term:escape` calls latent defect exposure "the only forbidden state". Emil ruled that the sentence stays in the record as he said it, holds only of a deviation never found by anyone, and is not a rule for the card, because "nobody can know 'never' in advance". +- **Claude, on C-2 (K-1).** Not an error in either party's reasoning. C-2 changed ruled wording without naming the supersession, and Claude asked. Emil ruled it a supersession of wording. +- **Claude, on its own gate reports at G1 (2026-11-18).** Claude reported `canon issue-refs --basis` as passing locally before the pushes of `49f19ed` and `7c1214e`. The container's clone was shallow, so the check covered no commit ("Refs checked on 0"). The workflow runs on GitHub, which fetch full history, did check them and passed (runs 384 and 385). Found at the filing, and fixed by unshallowing the local clone. + +## Emil at G2, verbatim + +Received at 13:41 Europe/Berlin (12:41 UTC) on 2026-11-18. + +> Hold at G2. I am not merging yet: "agent review" is a required check, and it fails on NU1004 +> until #83 lands. +> +> When I tell you #83 has merged: +> 1. Write P-4.6's result into predictions.md. +> 2. Push it. That re-runs the agent review workflow against the fixed main. +> 3. Name the new head to ratify, and report both required checks on it: +> "pipeline (devcontainer)" and "agent review". +> If the agent review then raises findings about this branch, report them and change nothing +> until I rule. +> +> Record two observations as successor items for #82, without changing the card: +> - the pull statement's "Observed" field passes only when the observer is inside the scope, +> while the worked example has a customer finding the fault; +> - the gap "specified, with nothing actual" names a state the card says is never read. +> +> Leave the DDD-address-11 warning as it is. + +| # | Time | Status | Reading | +|---|---|---|---| +| G2-1 | 13:41 | **Ruled** | Held at G2 until #83 lands. Then P-4.6's result is written, pushed, and the new head named with both required checks reported. Agent review findings are reported, not acted on | +| G2-2 | 13:41 | **Ruled** | Two observations recorded as successor items for #82 (`successor-items.md`, X-1 and X-2). The card is not changed | +| G2-3 | 13:41 | **Ruled** | `DDD-address-11`'s single-limb warning stays | + +## Emil on #82, verbatim + +Received at 13:46 Europe/Berlin (12:46 UTC) on 2026-11-18. + +> Thank you for catching the line in #82. It is corrected. +> +> #82 now lists what I ruled under #80: "actual" is never read directly; a deviation is found +> once a holder observes an outcome outside their tolerance; exposure nobody has observed is +> latent, which canon calls escaped, and finding it early is the work of assurance. The sentence +> "a bug that is never found has no impact" is removed from its "already ruled" list, with a +> correction note saying it was listed in error, that you caught it, and what I ruled at 13:26. +> X-1 and X-2 are added to #82 as successor items to take up there. +> +> Record this message in rulings.md with the time you receive it, and record the correction to +> #82 with its attribution: the error was Claude's in the planning conversation, and you caught +> it. Add it to the local commit. Do not push. +> +> Nothing else changes. Keep holding at G2 on 3d76c88. When I tell you #83 has merged, do as +> agreed: write P-4.6's result, push, name the new head, and report both required checks on it. + +| # | Time | Status | Reading | +|---|---|---|---| +| G2-4 | 13:46 | **Recorded**, done by Emil in #82 | #82's "already ruled under #80" list now reads: "actual" is never read directly; a deviation is found once a holder observes an outcome outside their tolerance; exposure nobody has observed is latent, which canon calls escaped, and finding it early is the work of assurance. "A bug that is never found has no impact" is removed from that list, with a correction note. X-1 and X-2 are added to #82 as successor items | +| G2-5 | 13:46 | **Ruled** | Still held at G2 on `3d76c88`. Nothing pushed until #83 has merged | + +## Correction, to #82 + +- **Claude in the planning conversation, caught by this session's Claude.** #82, written in the claude.ai planning conversation, listed "a bug that is never found has no impact" as already ruled under #80. That was an error: at 13:26 (K-3) Emil ruled that the sentence stays in the record as he said it and is not a rule for the card. This session caught it at 13:41 against `DDD-dec-122` and K-3. Emil corrected #82 at 13:46, with a correction note in the issue. diff --git a/meta/sessions/2026-11-18-act-map-scope-and-pull/steps-and-the-map.md b/meta/sessions/2026-11-18-act-map-scope-and-pull/steps-and-the-map.md new file mode 100644 index 0000000..8b34258 --- /dev/null +++ b/meta/sessions/2026-11-18-act-map-scope-and-pull/steps-and-the-map.md @@ -0,0 +1,51 @@ +# What steps 2 to 10 add to the act map + +Every one of steps 2 to 10 attaches to the act map, either to one of its parts or to a record kept beside it. None needs a part the map does not have. That is the evidence for S-1 in `proposals.md`: the act map as the index every later artefact attaches to. + +**Register: `[PROPOSED]`.** Steps 2 to 10 are `[PROPOSED]` and unfiled beyond one line each in the act address session's method draft. The "Value made visible" rows below are that line's value clause, verbatim, followed by one sentence on what the practitioner sees on the map. The second sentence is Claude's, `UNVERIFIED — Emil review`. + +**Weakest point.** Each attachment is read from one line of a draft nobody has ruled. A step whose content is ruled later may attach somewhere else, and its row here would then be wrong. + +## Where each step attaches + +The map's parts are those of `DDD-address-06` as proposed (P-1): scope, act sites, layers, dependencies and the index of determinables. Three records sit beside it: the occurrence record (step 9's), the derivation record (`DDD-address-02`'s "derivation layer") and the occupancy record (who occupies an address, kept out of the address by `DDD-ground-06`). + +| Step | What it adds | Attaches to | In the map or beside it | +|---|---|---|---| +| 2 Anchor authority | The top authority, and the role chain beneath it | The top acts and the composite acts below them: one authority per layer | In the map, on the layers | +| 3 Surface decisions | Each decision, anchored at the act site where it is resolved | The index: a determinable taken up becomes a decision. A decision with no act site is a candidate escape | In the map, on the index and the act sites | +| 4 Declare ground | Per decision, the provenance of its ground and its tick rate | The dependencies (whose ground reading carries what) and the index's rate of change | In the map, on the dependencies and the index | +| 5 Classify closure and set assurance | Per decision, which closure senses hold and the assurance its stakes need | The index's observed state: whether a verdict there is mechanical or judgement | In the map, on the index | +| 6 Allocate | An actor or arrangement at each address; new seams where decomposing pays | The occupancy record, beside each act site. A new seam adds a composite act, which is a layer | Beside the map, and in it for the layers | +| 7 Complete accountability | Attribution, principal, authority, stake and remediation per allocation | The occupancy record | Beside the map | +| 8 Derive forward | The ground requirements each act site pulls, and the components and slices they pull | The derivation record, by an edge from each pulling act site | Beside the map | +| 9 Execute and record | Each occurrence, verdict, correction and waiver | The occurrence record, each entry against exactly one act site with its time index. It fills the index's observed state | Beside the map | +| 10 Observe and supersede | Behaviour at fixed addresses; superseded decisions | The occurrence record, read at the two grains of `DDD-address-05`. A supersession changes the index's specified state, and can open a new pull | Beside the map, and in it on the index | + +Steps 6, 7, 8 and 9 attach beside the map, and steps 6 and 10 also change a part of it. No step needs an attachment the map lacks. + +## The "Value made visible" rows, proposed + +Each row replaces "Not filed beyond the quoted line" on the card, by a superseding section (`effects.md`). The value clause is quoted from the card's own "What is filed, verbatim" section. + +| Card | Proposed "Value made visible" row | +|---|---| +| [02](../../../method/02-anchor-authority.md) | "every indexed act has an owner, and acts cannot run when their principal leaves." On the map: an authority at every layer, so an act site whose layer has none shows as unowned | +| [03](../../../method/03-surface-decisions.md) | "the backlog becomes visible, and escape becomes countable per act." On the map: the index's determinables split into decisions, anchored at act sites, and those nobody took up | +| [04](../../../method/04-declare-ground.md) | "missing ground is found before build, not in production." On the map: a dependency whose ground has no declared provenance, and a determinable whose rate of change nobody declared | +| [05](../../../method/05-classify-closure-and-set-assurance.md) | "the practitioner knows where automation is safe and where judgment is load-bearing." On the map: which observed states can be read by a mechanical check and which need a person | +| [06](../../../method/06-allocate.md) | "an explicit map of where demand sits and who carries it." On the map: an occupant beside every act site, and the seams decomposition made | +| [07](../../../method/07-complete-accountability.md) | "an auditable answer to the compliance question." On the map: for every occupied act site, who answers for it and who is liable | +| [08](../../../method/08-derive-forward.md) | "traceability from act to decision to code, by construction." On the map: from any act site, the components it pulls, and from any component, every act site that pulls it | +| [09](../../../method/09-execute-and-record.md) | "evidence, not assertion." On the map: each act site's occurrences in causal order, and the observed state of each determinable, read from them | +| [10](../../../method/10-observe-and-supersede.md) | "escape is detected as it happens, and change is governed rather than absorbed." On the map: behaviour changing at a fixed address, and the pulls a supersession opens | + +All nine rows can be filled. None has to say it cannot be. + +## Amendment after G1 + +Appended on G1-Q9 (`rulings.md`). The rows above are left as written, and where they differ, this section governs. + +**Row 5 attaches closure to the act sites, not to the index's observed state.** The rows are `[PROPOSED]`. The reason Emil gave, "Closure is a property of an act's tolerance", is held at K-2: upstream `term:closure` defines a predicate as "closed for an arrangement". The card's row is written once K-2 is answered. + +**K-2, ruled 13:26.** Row 5 attaches closure to the act site's acceptance predicate, closed for the arrangement allocated there (`term:closure`). Emil withdrew the reason "closure is a property of an act's tolerance". Open, not resolved here: step 5 classifies closure before step 6 allocates the arrangement. The card's row is filed in that form. diff --git a/meta/sessions/2026-11-18-act-map-scope-and-pull/successor-items.md b/meta/sessions/2026-11-18-act-map-scope-and-pull/successor-items.md new file mode 100644 index 0000000..425eb90 --- /dev/null +++ b/meta/sessions/2026-11-18-act-map-scope-and-pull/successor-items.md @@ -0,0 +1,10 @@ +# Successor items — the act map, scope and pull, 2026-11-18 + +Two observations on step 1's first part, recorded for #82 on Emil's instruction of 13:41 (`rulings.md`). Neither is ruled. The card is not changed. + +| # | Observation | Where it shows | For | +|---|---|---|---| +| X-1 | The pull statement's "Observed" field passes only when the observer is inside the scope. The worked example has a customer finding the fault, and a customer is outside the scope | `method/01-map-and-index-acts.md`, the pull statement ("The observer is inside the scope") and the worked example's #103 gap row ("A customer who sees the line has also found it: expected against observed"); `DDD-address-11`'s region ("Observed is what someone inside the scope has seen"); `DDD-address-09`'s region ("any holder observes an outcome outside their tolerance") | #82 | +| X-2 | The gap "specified, with nothing actual" names a state the card says is never read | `method/01-map-and-index-acts.md`, the gaps table and the "Actual" row ("not read"); `DDD-address-09`'s region, which lists the gap; `DDD-address-11` ("actual is never read directly") | #82 | + +Also open, carried from this session and not for #82 alone: step 5 classifies closure before step 6 allocates the arrangement it is closed for (K-2). O-2 to O-5 and O-9 of 2026-11-17 stay open. diff --git a/meta/sessions/2026-11-18-pull-and-scope/bootstrap.md b/meta/sessions/2026-11-18-pull-and-scope/bootstrap.md new file mode 100644 index 0000000..071b45d --- /dev/null +++ b/meta/sessions/2026-11-18-pull-and-scope/bootstrap.md @@ -0,0 +1,15 @@ +# Bootstrap — the pull and the scope of a map, 2026-11-18 + +| Parameter | Value | +|---|---| +| Invocation | Interactive. A continuation of the claude.ai project conversation that ran the 2026-11-17 act map session. No issue is open. `prompt.md` carries Emil's statements verbatim | +| Principal | Emil | +| What this record is | **Rulings recorded when given.** No claim, decision, card or lifecycle file is changed. Nothing is filed until an issue is opened and its gates are ruled | +| Upstream pin | `v5.15.2` (`graph/upstream.yaml`), unchanged | +| Base | `66bc21f`, head of `claude/act-map-and-process-2026-11-17`, which waits at G2 and is not landed. `main` is at `84aa466`. The rulings here cite `DDD-address-06` to `-08` and `DDD-dec-122`, which exist only on that branch | +| Branch | `claude/pull-and-scope-2026-11-18`. Held locally until 09:18, because the repository is public and R-5 records a concept that is not Emil's or canon's. Pushed after R-6, which credits it to GitHub user `pjarnhus` | +| Model | Claude Fable 5.2 | + +## Decision ids + +Nothing is filed, so no id is taken. At 2026-11-17 20:26 UTC the highest decision id on any branch of either repository was `DDD-dec-122`, on the base branch. diff --git a/meta/sessions/2026-11-18-pull-and-scope/prompt.md b/meta/sessions/2026-11-18-pull-and-scope/prompt.md new file mode 100644 index 0000000..79e03c4 --- /dev/null +++ b/meta/sessions/2026-11-18-pull-and-scope/prompt.md @@ -0,0 +1,23 @@ +# The pull, the scope of a map, and the act map as the central artefact + +No issue is open for this work. Emil's statements below, verbatim, from the claude.ai project conversation on 2026-11-18 (Europe/Berlin), are the whole of the charter. A draft issue was written in the conversation at 07:23 and is not opened. + +> **07:20.** I think the act map should be the central modelling artifact for decision-driven-design - by creating it we get to work towards all other outcomes of the theory + +> **07:23.** Because whether we are applying DDD to software design, organization design or something else. We always need to start with mapping the acts and the decisions. Also i agree that we should start with North, only by declaring the determination pull can we properly map the acts and decisions + +> **08:43.** Now in order for us to start with the pull, we need a definition of it so that it can be reproduced by practicitioners + +> **08:50.** Name two outcomes that cannot both hold. If you cannot, it is not a determinable. - this should be two or more outcomes. - I think the most important part here is the scope of the map, by setting the scope of the map we decide when the stopping rule comes into effect. We could decide that the how is our world and then we would stop when we came outside of the how which is essential the SDLC + +> **08:57.** some issues here - if we started with the PDLC we wouldnt see the seam in the how because it wouldnt exist. The pull kinds still exist no matter where they come from because of their ground axes would be different - a bug and a feature cant share ground because ground for a feature is by necessity that it hasnt been implemented and a bug must have been implemented. - A friend of mine is working on a concept for the difference between specificed implementation, actual implementation, observeret implementation and expected implementation - which ties into our problem here + +> **09:08.** 1. yes - it must be else it cant be pull if we cant determine (a small decision driven design joke for you there) the ground it pulls from 2. the states are all that i have, but i completely agree with your mapping to the canon, this is exactly what we need in order create a process for our practicationers. This is kind a like looking at thermal dynamic laws, they also inspect a system and look at the input and output 3. scope must be the first else we cant place the pull, or we would end up far far from our area of influence + +> **09:18.** his github handle is pjarnhus lets start with that + +> **09:37.** you create the issues and i will run a session in claude code that actual changes the repo + +> **09:40.** lets just delete the branch for the old #79 work and then lets add all the new work into issue #80 - + +> **09:46.** agreed on your recommendation and accepted. Do it diff --git a/meta/sessions/2026-11-18-pull-and-scope/rulings.md b/meta/sessions/2026-11-18-pull-and-scope/rulings.md new file mode 100644 index 0000000..cc759cf --- /dev/null +++ b/meta/sessions/2026-11-18-pull-and-scope/rulings.md @@ -0,0 +1,53 @@ +# Rulings — the pull and the scope of a map, 2026-11-18 + +Emil's rulings and statements, in the order given. His words are in `prompt.md`, verbatim, by time. Nothing here is filed. + +| # | Time | Status | Emil's words (short form; full text in `prompt.md`) | Reading | +|---|---|---|---|---| +| E-1 | 07:20 | Direction given | "the act map should be the central modelling artifact for decision-driven-design" | Every application starts from the act map. Claude proposed central as an index that other artefacts point at, not as a container. Not ruled | +| E-2 | 07:23 | Direction given | "we should start with North, only by declaring the determination pull can we properly map the acts and decisions" | Step 1 starts by declaring north. Supported by `core/00` §4a at `v5.15.2`: "declaring the determinable space precedes any determination over it" (`DDD-ground-05`, projected). Step 1 is ruled for test and T2b is pre-registered with acts first, so this is a supersession. Not ruled | +| E-3 | 08:43 | Asked for | "we need a definition of it so that it can be reproduced by practicitioners" | Claude proposed: a pull is an unresolved determinable that the world resolves whether or not anyone acts, declared by a pull statement whose fields each carry a test, and found by a trace-back with a stopping rule | +| R-1 | 08:50 | **Ruled** | "this should be two or more outcomes" | The determinable's test reads: name two or more outcomes that cannot all hold together | +| R-2 | 08:50, 09:08 | **Ruled** | "by setting the scope of the map we decide when the stopping rule comes into effect"; "scope must be the first else we cant place the pull, or we would end up far far from our area of influence" | Scope is the first declaration of a map. Everything outside the scope is world to the map, including other people's decisions. The trace-back stops where the answer lies outside the scope. Emil's example: with the How as the scope, the trace stops at the edge of the SDLC | +| R-3 | 08:57 | **Correction by Emil** | "if we started with the PDLC we wouldnt see the seam in the how because it wouldnt exist" | A seam exists only where a map splits the work into scopes. Claude's check is corrected: a pull of a narrow scope appears in a wider map as a dependency (`DDD-address-01`), and as a seam only if the wider map also declares that scope | +| R-4 | 08:57, 09:08 | **Ruled** | "a bug and a feature cant share ground because ground for a feature is by necessity that it hasnt been implemented and a bug must have been implemented"; "yes - it must be else it cant be pull if we cant determine […] the ground it pulls from" | A kind of demand is a position on a declared ground axis. Kinds differ by ground, not by where the demand came from. Consistent with `DDD-flow-01`, which defines the PDLC's kinds relative to the decisions in force | +| R-5 | 09:08 | **Ruled**, on four names only | "the states are all that i have, but i completely agree with your mapping to the canon, this is exactly what we need in order create a process for our practicationers" | The four states map to canon as in the table below, and a pull is a gap between two of them. The concept is a friend of Emil's. Only its four names are known. See "The four states" | +| E-4 | 09:08 | Statement | "This is kind a like looking at thermal dynamic laws, they also inspect a system and look at the input and output" | The pulls of a scope can be found by inspecting its edge, before any act inside it is mapped. Claude's reading, `UNVERIFIED — Emil review` | +| R-6 | 09:18 | **Ruled** | "his github handle is pjarnhus lets start with that" | Asked: "May the four states appear in the public repository, and how should your friend be credited?" The four states are credited to GitHub user `pjarnhus`. Claude reads "lets start with that" as leave to record the concept here under that credit, and as a first form of credit that may change. `UNVERIFIED — Emil review` | +| R-7 | 09:37, 09:40 | **Ruled** | "you create the issues and i will run a session in claude code that actual changes the repo"; "lets just delete the branch for the old #79 work and then lets add all the new work into issue #80" | One issue, #80, carries all of the work of both days. #79 is closed and folded in, and #78 is closed without landing. The repository is changed by a Claude Code session of Emil's, not by this conversation. The branch `claude/act-map-and-process-2026-11-17` is to be deleted; this session was refused the deletion, so Emil deletes it. This branch is the archive until #80's branch is pushed | +| R-8 | 09:46 | **Ruled** | "agreed on your recommendation and accepted. Do it" | On Claude's recommendation of 09:44. The claims filed under #80 keep their ids: `DDD-address-06` (the act map), `-07` (a second act map), `-08` (a process and north) and `DDD-dec-122`, each amended before landing, as `DDD-flow-03` was. New claims take `DDD-address-09` onwards. Each amended claim says in its notes that it was first drafted on 2026-11-17 on a branch that did not land and was amended on the rulings of 2026-11-18. The archive branch stays until #80's branch is pushed, so the ids show as taken. Claude reads "Do it" as: record this and put it in #80. The filing itself is #80's session's. `UNVERIFIED — Emil review` | + +## Corrections + +- **Claude, on the gate (2026-11-17 and 2026-11-18).** Claude reported the gate as passing on `python3 scripts/ci.py` alone. The `validate` workflow also runs `canon issue-refs --basis`, which refuses every commit without a `Refs #N` line. Every push of the 2026-11-17 session failed it, and Claude did not look at the workflow runs. Point 7 of that session's R-7 ("no issue number on the commits") cannot stand with the pipeline, and is superseded by #80. +- **Claude, on R-3.** Claude proposed that a pull of a narrow scope "must show up in any wider map as a seam". Wrong: the wider map need not split the work. Corrected by Emil. +- **Claude, on R-4.** Claude proposed an origin field on the pull statement (the world moving, or another scope's act) to keep failure demand visible in a narrow scope. Withdrawn: the kind is read from the ground position. +- **Claude, on R-2.** Claude wrote that "the pull itself is in the world, but where the trace stops depends on whose arrangement is being mapped". Emil's ruling is stronger: the scope defines what counts as world. + +## The four states + +The concept is not Claude's and not canon's. **Credit: GitHub user `pjarnhus`** (R-6), a friend of Emil's, who is working on the difference between four states of an implementation. The work is not published, and nothing here is his wording beyond the four names. Emil has the four names and no more. The mapping is Claude's, made from the names, and Emil agreed with it (R-5). + +| State | Nearest thing in canon at `v5.15.2` | +|---|---| +| Specified | The governing decisions in force | +| Actual | The outcome: "the determinate as it lands in the world" (`term:outcome`) | +| Observed | The verdict: "the same determinate as assessed by a declared predicate" (`term:outcome`) | +| Expected | The tolerance held by whoever bears the outcome | + +| Gap | What it is | +|---|---| +| Expected, with nothing specified | A need nobody has taken up: a feature | +| Specified, with nothing actual | An approved change not yet built | +| Specified against actual | A deviation: a bug, which needs an implementation to exist | +| Actual against observed | A deviation nobody has seen, or an alarm with nothing behind it | +| Expected against specified | The specification is wrong | + +`UNVERIFIED — Emil review`, and open: +- The mapping is built from four words. The friend's own definitions may differ. +- "Implementation" is a software word. The four states have not been tested as states of any determinable. +- "Expected" has one holder per stakeholder, so it is not one state. +- "Observed" depends on scope: it is what someone inside the scope has seen. +- "Actual" is never read directly. The world renders outcomes and not verdicts, so a practitioner can fill "observed" and can only declare "actual" unknown. The fourth gap cannot be inspected, only suspected. +- Own-ground demand, such as a CVE, is about the arrangement and not about the product's implementation. Where it sits among the gaps is not worked out. +- The credit is a handle and no more (R-6). Whether `pjarnhus` has agreed to the concept appearing here, and how he wants it cited once it is published, is not recorded. diff --git a/meta/sessions/README.md b/meta/sessions/README.md index 75f7bba..23237c9 100644 --- a/meta/sessions/README.md +++ b/meta/sessions/README.md @@ -54,4 +54,5 @@ session's later work exceeds the committed prompt, the excess is visible as exce | `2026-11-08-specification-feasibility/` | Rulings filed, gated — specification feasibility, from a claude.ai conversation (F-1 to F-12): every dependency on ground has a What layer and a How layer, the How decides by three loose-coupling conditions, any granted actor vets a pinned What change and product-cli records the decision (`DDD-dec-116`); vetted ground becomes a ground-sourced bounded context, admitted per system for Deciders, Projectors and Views, with permission the data owner's; product-cli and the framework carry grants and roles, superseding `[D1.11]` (`DDD-dec-117`). Not applied; S-1 to S-8. Issue #62 | | `2026-11-08-demand-flow/` | Rulings filed, gated — map the flow of demand, from the claude.ai conversation's Rounds 23 to 25: a method step after step 1 (`method/01b-map-the-flow-of-demand.md`, `[PROPOSED]`), general, with each process declaring its kinds of demand and the PDLC's new, owed and own-ground as its worked example and starting point; four PDLC-scoped claims in a new area, `DDD-flow-01` to `-04`, projected; `DDD-dec-118`, adding to `DDD-dec-114`; the acme shop's #101, #103 and #104 traced as graph data checked against the act map. Nothing upstream. Round 26 (2026-11-10) closed R-1: sighting separated from entry, only the What owner says whether a defect is a bug. Open: numbering (N-1), promotion (P-1), the two examples (E-1). Issue #67 | | `2026-11-10-release-v0.7.0/` | Release descriptor, gated. `releases/v0.7.0.yaml` is MINOR under `DDD-dec-82`: step 1b's card was added (`DDD-dec-118`), and no step's register moved (F-1). Format 2, `prerelease: true` for Emil to confirm (F-3). Its `CHANGELOG.md` section is from `canon changelog`, and `canon ratify` fills the ratified head. The prompt calls the PDLC and SDLC specifications `[PROPOSED]`, while the repository says they are ruled (Round 28). The summary follows the repository, and Emil is to settle it (F-2). Not landed and not cut. Issue #69 | +| `2026-11-18-act-map-scope-and-pull/` | Interactive, gated — the act map, scope and pull (#80). Carries the two session records of 2026-11-17 and 2026-11-18 from the archive branch at `47cf4cc`, unreworded. Filed (`DDD-dec-122`): the act map with its scope and index of determinables (`DDD-address-06`); a second act map with a second scope or a second declaration of axes (`-07`); a process as a named view, north as the determinable of a pull (`-08`); the pull as a gap between two of four states (`-09`); kinds as ground positions (`-10`); the four states, credited to GitHub user pjarnhus (`-11`). Step 1 starts from scope and pull, a `[PROPOSED]` first part; steps 2 to 10 say what they add to the act map; step 1b reads kinds as positions. Superseding sections, nothing reworded. Three conflicts with canon or earlier rulings caught before filing (K-1 to K-3). Issue #80 | | `2026-11-18-canon-lock-file/` | Interactive, gated — the `agent review` check builds again (#83). NU1004 reproduced: `IsAotCompatible` makes the SDK reference `Microsoft.NET.ILLink.Tasks` at its bundled runtime (10.0.15 in SDK 10.0.404), the lock file holds 10.0.14, and the runner sets `CI`. Corrected: `validate` passes because `CI` is unset in the devcontainer, not because of an older SDK; and agent review builds main's canon, so no pull request can pass it until main builds (D-4). Proposals A (regenerate), B (exact SDK pin), C (pin the ILLink pack); G1 open. Issue #83 | diff --git a/method/01-map-and-index-acts.md b/method/01-map-and-index-acts.md index 5ec5e02..f0c02c4 100644 --- a/method/01-map-and-index-acts.md +++ b/method/01-map-and-index-acts.md @@ -12,6 +12,86 @@ nothing on this card is reported, and the step should be tried, not relied on (` region). A map made with the wrong axes looks complete and is not, and the step itself cannot tell you which axes you missed. +## First part: declare the scope, then find the pulls `[PROPOSED]` + +**This section supersedes, where they differ, the lines below it** (`DDD-dec-122`, ruled 2026-11-18). Those lines stay as written. Three are superseded: +- "Map the acts the work consists of, forward from the domain and the request" becomes: forward from the entry acts this first part names. +- The "Ground in" row gains the scope and the pull statements. +- The "Artefact out" row's "a tree of occurrences" becomes the act map of `DDD-address-06`: act sites at their addresses, the composite acts that layer them, the dependencies between them and the index of determinables, with occurrences recorded beside it. + +**Register.** This first part is `[PROPOSED]` inside a card ruled for test. Its claims, `DDD-address-06`, `-09`, `-10` and `-11`, are projected. The rest of the card keeps its register. + +**Why it comes first.** A pull cannot be placed until the map's scope says what is world, and acts cannot be mapped until the pulls say where demand enters. Declaring the determinable space precedes determination over it (`DDD-ground-05`). + +A **pull** is a gap between two of four states of something that can turn out more than one way: specified, actual, observed and expected (`DDD-address-09`, `DDD-address-11`). The four state names are GitHub user pjarnhus's. The mapping to canon is ours. + +### The five parts + +| # | Part | The question | Rests on | +|---|---|---|---| +| 1 | Scope | **Which decisions can the people doing this work make without asking anyone else?** List them. That list is the scope. Every other decision, and everything nobody decides, is the world. The world still reaches you, through what comes in across the edge | `DDD-address-06` | +| 2 | The determinables at the edge | What comes into the work and what goes out of it? For each, name two or more ways it could turn out that cannot all be true together. If you cannot, it is not a determinable | `DDD-address-09` | +| 3 | The three readable states | What was decided about it (specified)? What has someone inside the scope seen (observed)? What does each person who bears the outcome expect (expected)? Write "not read" under actual: you only ever see what someone observed | `DDD-address-11` | +| 4 | The gaps | Which two states disagree? Read the kind from the map's declared axis. Start from **built or not built** unless your map declares another | `DDD-address-09`, `DDD-address-10` | +| 5 | The trace to an entry act | Follow the gap back. Stop where someone outside the scope would have to answer. Which act inside the scope takes it in? | `DDD-address-09` | + +### The pull statement + +One per pull. Each field has a check that passes or fails. + +| Field | Holds | Passes when | +|---|---|---| +| Scope | The decisions inside, in plain words | Every decision named is marked inside or outside, once | +| Edge crossing | What crosses in or out, who produces it, who reads it | Exactly one side is inside | +| Determinable | What can turn out more than one way | It crosses the edge | +| Outcomes | Two or more | No two can hold together (a judgement) | +| Specified | The decision in force, cited, or "none" | A citation that resolves, or "none" | +| Observed | What was seen, by whom, where, or "not observed" | The observer is inside the scope | +| Expected | One row per holder who bears the outcome | Each row names its holder | +| Actual | "not read" | It says "not read" | +| Gap | The two states that disagree | Both are filled, or the gap allows "none" | +| Kind | The position on the map's declared kind axis | The axis is named and the position is one of its values. The source decides nothing | +| Rate of change | How often time makes it turn out again, with a unit, or "no unit known" | It is stated | +| Trace and stop | The chain followed back, and where it stops | Every link but the last is inside. The last is world, or the chain ends inside at the cause (a judgement) | +| Entry act | The first act inside the scope that reads the pull | It is inside, and the trace reaches it | + +**Gaps, as examples, not a full list.** + +| Gap | What it is | +|---|---| +| Expected, with nothing specified | A need nobody has taken up | +| Specified, with nothing actual | An approved change not yet built | +| Specified against actual | A bug. Readable only through its stand-in, specified against observed | +| Actual against observed | A deviation nobody has seen. It can be suspected, never inspected | +| Expected against specified | The specification is wrong | +| Expected against observed | The software matches its specification, and the holder still cannot do the thing | + +### What this part finds, and what it does not + +- **It finds a deviation once a holder observes an outcome outside their tolerance**, whether or not they trace it to the software. That is "expected against observed", found from the expected side. +- **Exposure that nobody has observed is latent**, which canon calls escaped (upstream `term:escape`). This part does not find it. +- **Finding it before it lands is the work of assurance**, in [step 5](05-classify-closure-and-set-assurance.md), not of finding pulls. + +### Worked example, in brief + +The acme shop of [`applications/pdlc/`](../applications/pdlc/), invented, with the team that builds and runs its software as the scope. Its people can decide how a change is planned, built, checked, landed and deployed. They cannot decide what the shop must do, which work comes first, or when a feature is switched on. Issue #103: reorder adds an out-of-stock line to the basket. + +| Field | #103 | +|---|---| +| Determinable and outcomes | Whether reorder respects the stock invariant: never adds an out-of-stock line, or adds one | +| Specified | The Place order Decider's stock invariant, decided outside the scope | +| Observed | The out-of-stock line in a basket, on the team's telemetry | +| Expected | Customers and the people who own what the shop must do: never adds one | +| Actual | Not read | +| Gap | Specified against observed, the stand-in for a bug. A customer who sees the line has also found it: expected against observed | +| Kind | Built | +| Trace and stop | The request to fix reorder ← whether this is a bug or the specification is wrong. That answer lies outside the scope. Stop | +| Entry act | Plan the realisation (o1b in the acme map) | + +Under a wider scope, the whole product lifecycle, the same item's trace runs past that answer, and it enters at the classification act (oC), as `DDD-flow-03` says. The full example, with #101 and #104, is in the session record [`pull-procedure.md`](../meta/sessions/2026-11-18-act-map-scope-and-pull/pull-procedure.md). + +**The junior-engineer desk check was a reading, not a run.** Claude read this part as a junior engineer would meet it, and the session record lists what failed and how it was fixed. Nobody has run the junior-engineer test on this part. + ## What the step does Map the acts the work consists of, forward from the domain and the request, never backward from diff --git a/method/01b-map-the-flow-of-demand.md b/method/01b-map-the-flow-of-demand.md index 14a6265..f84f7cb 100644 --- a/method/01b-map-the-flow-of-demand.md +++ b/method/01b-map-the-flow-of-demand.md @@ -18,6 +18,23 @@ process has the same three. One process is the whole of the evidence (Round 24: have enough evidence to say if the three kinds of demand entities are general enough for all processes"). Every use of this step on a second process is a test of them. +## Kinds as ground positions, and a process `[PROPOSED]` + +**This section supersedes, where they differ, the lines below it** (`DDD-dec-122`, ruled 2026-11-18, which amends `DDD-dec-118`). Those lines stay as written. + +| Card text below | Read as | Basis | +|---|---|---| +| "each process declares its own kinds of demand" | Each map declares its kinds, as positions on a declared ground axis. The method offers **built or not built** as the starting axis, as it offers the PDLC's kinds as the starting set | `DDD-address-10`; `DDD-address-08`: a process is not an object of the model, so it does not declare | +| Part 1, "Map acts forward from the request" | Map acts forward from the entry acts named by step 1's first part, the pull procedure | `DDD-dec-122` | +| Part 6, "Classify each source of pull into the kinds of demand the process declares" | Read each pull's kind from its position on the map's declared axis. The source is kept as where the item was sighted, and decides nothing about its kind | `DDD-address-10` | +| "a process" | A human understanding of work. Its translation into the model is a named view of one act map, fixed by one or more norths, each the determinable of a pull, and optionally bounded by composite acts | `DDD-address-08` | + +**What stands.** The PDLC's three kinds, new, owed and own-ground, and their one-line tests. `DDD-flow-01` to `-04`, projected and scoped to the PDLC: `DDD-flow-01` already tests each kind relative to the decisions in force, which is a reading of ground. The step's place after step 1. "Sighting is not entry". + +**Own-ground demand on "built or not built".** On that axis alone, the acme shop's #104 (own-ground) sits with #103 (owed). A second axis, "whose determinable" (the product's or the arrangement's), would tell them apart. It is `[PROPOSED]`, with own-ground demand as its only evidence. + +**The field report** gains two columns, beside the table under "Record the declaration": the axis your kinds are read on, and each kind's position on it. The promotion item (P-1) then asks whether a second map's axis and positions match the PDLC's. + ## What the step does Step 1 maps acts, one occurrence at a time. It does not say how demand arrives in a running diff --git a/method/02-anchor-authority.md b/method/02-anchor-authority.md index f71bc6e..a1bc4cf 100644 --- a/method/02-anchor-authority.md +++ b/method/02-anchor-authority.md @@ -27,6 +27,17 @@ as a method step yet. | Exit check | Not filed | | Principal | Not filed | +## What this step adds to the act map `[PROPOSED]` + +**This section supersedes, where they differ, the step card above** (`DDD-dec-122`, ruled 2026-11-18). The card stays as written. Its "Value made visible" row, "Not filed beyond the quoted line", now reads: + +| Field | Content | +|---|---| +| Value made visible | "every indexed act has an owner, and acts cannot run when their principal leaves." On the act map: an authority at every layer, so an act site whose layer has none shows as unowned | +| Attaches to | The top acts and the composite acts below them: one authority per layer. In the map, on the layers | + +The act map is defined in `DDD-address-06`. The value clause is the draft's, verbatim. The rest of the row is `[PROPOSED]`. + The default falsifier for any method step, also from the draft and also `[PROPOSED]`: "practitioners perform the step and cannot name the value it gave them." diff --git a/method/03-surface-decisions.md b/method/03-surface-decisions.md index bae9d63..519144b 100644 --- a/method/03-surface-decisions.md +++ b/method/03-surface-decisions.md @@ -27,6 +27,17 @@ as a method step yet. | Exit check | Not filed | | Principal | Not filed | +## What this step adds to the act map `[PROPOSED]` + +**This section supersedes, where they differ, the step card above** (`DDD-dec-122`, ruled 2026-11-18). The card stays as written. Its "Value made visible" row, "Not filed beyond the quoted line", now reads: + +| Field | Content | +|---|---| +| Value made visible | "the backlog becomes visible, and escape becomes countable per act." On the act map: the index's determinables split into decisions, anchored at act sites, and those nobody took up | +| Attaches to | The index of determinables, where a determinable taken up becomes a decision, and the act site where it is resolved. A decision with no act site is a candidate escape. In the map | + +The act map is defined in `DDD-address-06`. The value clause is the draft's, verbatim. The rest of the row is `[PROPOSED]`. + The default falsifier for any method step, also from the draft and also `[PROPOSED]`: "practitioners perform the step and cannot name the value it gave them." diff --git a/method/04-declare-ground.md b/method/04-declare-ground.md index c0981a3..740930d 100644 --- a/method/04-declare-ground.md +++ b/method/04-declare-ground.md @@ -27,6 +27,17 @@ as a method step yet. | Exit check | Not filed | | Principal | Not filed | +## What this step adds to the act map `[PROPOSED]` + +**This section supersedes, where they differ, the step card above** (`DDD-dec-122`, ruled 2026-11-18). The card stays as written. Its "Value made visible" row, "Not filed beyond the quoted line", now reads: + +| Field | Content | +|---|---| +| Value made visible | "missing ground is found before build, not in production." On the act map: a dependency whose ground has no declared provenance, and a determinable whose rate of change nobody declared | +| Attaches to | The dependencies, and the index's rate of change. In the map | + +The act map is defined in `DDD-address-06`. The value clause is the draft's, verbatim. The rest of the row is `[PROPOSED]`. + The default falsifier for any method step, also from the draft and also `[PROPOSED]`: "practitioners perform the step and cannot name the value it gave them." diff --git a/method/05-classify-closure-and-set-assurance.md b/method/05-classify-closure-and-set-assurance.md index 24d6737..6982cf1 100644 --- a/method/05-classify-closure-and-set-assurance.md +++ b/method/05-classify-closure-and-set-assurance.md @@ -27,6 +27,17 @@ as a method step yet. | Exit check | Not filed | | Principal | Not filed | +## What this step adds to the act map `[PROPOSED]` + +**This section supersedes, where they differ, the step card above** (`DDD-dec-122`, ruled 2026-11-18). The card stays as written. Its "Value made visible" row, "Not filed beyond the quoted line", now reads: + +| Field | Content | +|---|---| +| Value made visible | "the practitioner knows where automation is safe and where judgment is load-bearing." On the act map: at each act site, whether its acceptance predicate is closed for the arrangement allocated there, so where a verdict can be a mechanical check and where it needs a person | +| Attaches to | The act site's acceptance predicate, closed for the arrangement allocated there (upstream `term:closure`). In the map, on the act sites. **Open, not resolved here:** this step classifies closure before step 6 allocates the arrangement it is closed for | + +The act map is defined in `DDD-address-06`. The value clause is the draft's, verbatim. The rest of the row is `[PROPOSED]`. + The default falsifier for any method step, also from the draft and also `[PROPOSED]`: "practitioners perform the step and cannot name the value it gave them." diff --git a/method/06-allocate.md b/method/06-allocate.md index 513a2d3..61f05c8 100644 --- a/method/06-allocate.md +++ b/method/06-allocate.md @@ -27,6 +27,17 @@ as a method step yet. | Exit check | Not filed | | Principal | Not filed | +## What this step adds to the act map `[PROPOSED]` + +**This section supersedes, where they differ, the step card above** (`DDD-dec-122`, ruled 2026-11-18). The card stays as written. Its "Value made visible" row, "Not filed beyond the quoted line", now reads: + +| Field | Content | +|---|---| +| Value made visible | "an explicit map of where demand sits and who carries it." On the act map: an occupant beside every act site, and the seams decomposition made | +| Attaches to | The occupancy record, beside each act site, since the occupant is never part of the address (`DDD-ground-06`). A new seam adds a composite act, which is a layer. Beside the map, and in it for the layers | + +The act map is defined in `DDD-address-06`. The value clause is the draft's, verbatim. The rest of the row is `[PROPOSED]`. + The default falsifier for any method step, also from the draft and also `[PROPOSED]`: "practitioners perform the step and cannot name the value it gave them." diff --git a/method/07-complete-accountability.md b/method/07-complete-accountability.md index 029dff1..e84bd97 100644 --- a/method/07-complete-accountability.md +++ b/method/07-complete-accountability.md @@ -27,6 +27,17 @@ as a method step yet. | Exit check | Not filed | | Principal | Not filed | +## What this step adds to the act map `[PROPOSED]` + +**This section supersedes, where they differ, the step card above** (`DDD-dec-122`, ruled 2026-11-18). The card stays as written. Its "Value made visible" row, "Not filed beyond the quoted line", now reads: + +| Field | Content | +|---|---| +| Value made visible | "an auditable answer to the compliance question." On the act map: for every occupied act site, who answers for it and who is liable | +| Attaches to | The occupancy record. Beside the map | + +The act map is defined in `DDD-address-06`. The value clause is the draft's, verbatim. The rest of the row is `[PROPOSED]`. + The default falsifier for any method step, also from the draft and also `[PROPOSED]`: "practitioners perform the step and cannot name the value it gave them." diff --git a/method/08-derive-forward.md b/method/08-derive-forward.md index 3a6676c..2c83068 100644 --- a/method/08-derive-forward.md +++ b/method/08-derive-forward.md @@ -27,6 +27,17 @@ as a method step yet. | Exit check | Not filed | | Principal | Not filed | +## What this step adds to the act map `[PROPOSED]` + +**This section supersedes, where they differ, the step card above** (`DDD-dec-122`, ruled 2026-11-18). The card stays as written. Its "Value made visible" row, "Not filed beyond the quoted line", now reads: + +| Field | Content | +|---|---| +| Value made visible | "traceability from act to decision to code, by construction." On the act map: from any act site, the components it pulls, and from any component, every act site that pulls it | +| Attaches to | The derivation record, by an edge from each pulling act site (`DDD-address-02`). Beside the map | + +The act map is defined in `DDD-address-06`. The value clause is the draft's, verbatim. The rest of the row is `[PROPOSED]`. + The default falsifier for any method step, also from the draft and also `[PROPOSED]`: "practitioners perform the step and cannot name the value it gave them." diff --git a/method/09-execute-and-record.md b/method/09-execute-and-record.md index 7402ee2..03b3967 100644 --- a/method/09-execute-and-record.md +++ b/method/09-execute-and-record.md @@ -27,6 +27,17 @@ as a method step yet. | Exit check | Not filed | | Principal | Not filed | +## What this step adds to the act map `[PROPOSED]` + +**This section supersedes, where they differ, the step card above** (`DDD-dec-122`, ruled 2026-11-18). The card stays as written. Its "Value made visible" row, "Not filed beyond the quoted line", now reads: + +| Field | Content | +|---|---| +| Value made visible | "evidence, not assertion." On the act map: each act site's occurrences in causal order, and the observed state of each determinable, read from them | +| Attaches to | The occurrence record, each entry against exactly one act site with its time index (`DDD-address-01`). It fills the index's observed state. Beside the map | + +The act map is defined in `DDD-address-06`. The value clause is the draft's, verbatim. The rest of the row is `[PROPOSED]`. + The default falsifier for any method step, also from the draft and also `[PROPOSED]`: "practitioners perform the step and cannot name the value it gave them." diff --git a/method/10-observe-and-supersede.md b/method/10-observe-and-supersede.md index d9e5f63..5c4557b 100644 --- a/method/10-observe-and-supersede.md +++ b/method/10-observe-and-supersede.md @@ -27,6 +27,17 @@ as a method step yet. | Exit check | Not filed | | Principal | Not filed | +## What this step adds to the act map `[PROPOSED]` + +**This section supersedes, where they differ, the step card above** (`DDD-dec-122`, ruled 2026-11-18). The card stays as written. Its "Value made visible" row, "Not filed beyond the quoted line", now reads: + +| Field | Content | +|---|---| +| Value made visible | "escape is detected as it happens, and change is governed rather than absorbed." On the act map: behaviour changing at a fixed address, and the pulls a supersession opens | +| Attaches to | The occurrence record, read at the two grains of `DDD-address-05`. A supersession changes the index's specified state, and can open a new pull. Beside the map, and in it on the index | + +The act map is defined in `DDD-address-06`. The value clause is the draft's, verbatim. The rest of the row is `[PROPOSED]`. + The default falsifier for any method step, also from the draft and also `[PROPOSED]`: "practitioners perform the step and cannot name the value it gave them."