Skip to content

SPEC 16: fixes against one base, any set of them - #23

Draft
mklarmann wants to merge 1 commit into
step-rangesfrom
fixes-format
Draft

mklarmann wants to merge 1 commit into
step-rangesfrom
fixes-format

Conversation

@mklarmann

@mklarmann mklarmann commented Nov 17, 2026 •

Copy link
Copy Markdown
Collaborator

Summary

A proposal, as text only: SPEC section 16, fixes against one base. A chain is to hold hundreds of fixes, and a reader can choose any set of them. Manuel decided the shape on 18 November 2026: one base, with each fix measured on its own, in a file of its own. A set that the file cannot show to be exact is still shown, marked as estimated.

chain file (unchanged)      published → … → base → … → release
fixes file (new)            of: {file, title, base}
├── items, base             every item, base result per indicator
├── fixes                   a step's fields + effect (every change from base, sparse) + counts
├── interactions            {checked, method, by, pairs: [{fixes: [a, b], reason}]}
└── built                   sets the producer built (e.g. the release with every fix)
result of a set = base + Σ effects
label           = built      if the set is one fix or a built set
                  exact      if interactions were checked and no two of its fixes are a listed pair
                  estimated  otherwise: marked on every cell and in every count
tolerance       = largest of its fixes'
noise           = sum of its fixes'
verdict         = none (each claim is judged in its own fix, against the base)

What the validator will hold:

  • A built set must agree with the sum of its effects wherever the file calls its fixes independent. So a false claim of independence is refused, not drawn.
  • The release built with every fix tests every pair at once.

Evidence

Text only. On #22, 215 Python tests and 82 node tests pass, unchanged. Acceptance cases F1–F8 (section 16.11) are written to become tests before the code.

Merge Danger

Door: two-way. A new section and six glossary lines. No format, code or test changes.

Blast radius: none for readers of sentier-diagnostics/1. Once agreed, it decides the shape of a new format, sentier-diagnostics-fixes/1, and the work in sentier-bafu (the producer and the page).

Based on #22 (step-ranges): noise per state and the matrix's "=", which #24 builds on. This PR stays the text Manuel approved. #24 changes it where the validator decided:

  • a pair holds no text: {fixes, kind, reach, largest};
  • requires, added items and a built set's state;
  • noise per build (§5.3): a fix's comparison takes its build's noise plus that of the build it was measured against, and a set takes the sum of both over its fixes;
  • sentier-bafu's producer review: the size of a pair and "no size", the route an effect is measured on, and exact reading.

#24 also adds the pointer from section 3.

Reviewer: Manuel, for the text. Laurenz is asked only when the whole stack is done.

🤖 Generated with Claude Code

mklarmann added a commit that referenced this pull request Nov 17, 2026
The page no longer adds up the differences of steps apart. A step's difference
was measured with every step before it applied, and an LCA result is not
linear in the technosphere, so a sum would use differences measured in a state
the set never had. Any set of fixes belongs to a file of fixes measured on one
base (SPEC section 16, proposed in #23).

On a chain the page compares one step, a run of consecutive steps, or the
whole way. A run is judged from the state before its first step and the state
after its last, with the largest tolerance and the largest measured noise of
its steps. Steps apart, from an address, are refused and the reason is said.

The strip and the list choose one step, or with Shift (or Cmd or Ctrl) a run;
a group heading chooses its run. The ticks and "the steps where something
moved" are gone, since both made steps apart.

The page's port of rule.judge is now held to 813 pairs on every edge of the
rule, written by tests/test_boundaries.py: the tolerance, the noise, the floor,
a sign crossing, ten times, every band edge, factors too large for a float and
subnormal results, mirrored and below zero. Probes that turn one comparison of
the port from <= to < are caught.

together.py judges a run (first, last), and tests/fixtures/together.json holds
every run of the local example.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@mklarmann
mklarmann changed the base branch from page-review to dashboard-page November 17, 2026 05:24
@mklarmann
mklarmann changed the base branch from dashboard-page to first-version November 17, 2026 05:36
…ct or estimated

Manuel chose one base with fixes measured on their own over a chain of hundreds of
states, a file of fixes of its own, and estimated sets shown but marked (18 November
2026). This section says what that file holds, when a sum of effects is the result,
how a set is judged and labelled, how interactions are declared and held to built
sets, what the page does and what the validator will refuse. No code yet: the
validator follows once the text is agreed.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@mklarmann
mklarmann changed the base branch from first-version to step-ranges November 17, 2026 05:59
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant