Reduce complete publication preflight time by 34.2% (no new κ)

research/joint-source-ancestry/BENCHMARK.json

@@ -0,0 +1,181 @@
+{
+  "base_commit": "721c03822c421b7f7fcbb2598993b947c4d88419",
+  "baseline_source_sha256": "63026ec3805f2a3236cc071fc78cf1c54bc6528c43e2f3d82548dc68bf346bce",
+  "changed_source_sha256": "31a5d71c96e2b2e5bd1a81cc2210604410ad23477cb2f2d58c729530bf39dbf9",
+  "failures": 0,
+  "fixture": "research/e8-d0-root-swaps",
+  "git": "git version 2.54.0 (Apple Git-157)",
+  "load_average": [
+    3.5625,
+    3.7978515625,
+    3.55908203125
+  ],
+  "logical_cpus": 18,
+  "pairs": 7,
+  "platform": "macOS-27.0.1-arm64-arm-64bit",
+  "python": "3.12.15 (main, Oct 14 2026, 00:55:37) [Clang 23.1.3 ]",
+  "receipt": {
+    "commit": "721c03822c421b7f7fcbb2598993b947c4d88419",
+    "files": 14,
+    "package": "research/e8-d0-root-swaps",
+    "source_history_commits": [
+      "08e134b4a0c7f04869dae703f2a2e688e53c44f0",
+      "18f7d4da808edb82d32e2081f869b8c993eca891",
+      "41826461eaceb25b77c026148b50b884e3c9780e",
+      "4c32ac3b88d63ad06d2b30e526aa07a12929db36",
+      "997e6a9ca7b4f95d50e366f61412fd640d72a88a"
+    ]
+  },
+  "receipt_sha256": "92528be17fc5781fc54e65c294e9a607aa19fe9c6eddf5249ce07cf71906f796",
+  "samples": [
+    {
+      "pair": 0,
+      "returncode": 0,
+      "seconds": 0.24568341695703566,
+      "side": "baseline"
+    },
+    {
+      "pair": 0,
+      "returncode": 0,
+      "seconds": 0.15555229200981557,
+      "side": "changed"
+    },
+    {
+      "pair": 1,
+      "returncode": 0,
+      "seconds": 0.17523508286103606,
+      "side": "changed"
+    },
+    {
+      "pair": 1,
+      "returncode": 0,
+      "seconds": 0.23902325006201863,
+      "side": "baseline"
+    },
+    {
+      "pair": 2,
+      "returncode": 0,
+      "seconds": 0.24392245803028345,
+      "side": "baseline"
+    },
+    {
+      "pair": 2,
+      "returncode": 0,
+      "seconds": 0.1694880418945104,
+      "side": "changed"
+    },
+    {
+      "pair": 3,
+      "returncode": 0,
+      "seconds": 0.20432179211638868,
+      "side": "changed"
+    },
+    {
+      "pair": 3,
+      "returncode": 0,
+      "seconds": 0.24606208316981792,
+      "side": "baseline"
+    },
+    {
+      "pair": 4,
+      "returncode": 0,
+      "seconds": 0.2425744580104947,
+      "side": "baseline"
+    },
+    {
+      "pair": 4,
+      "returncode": 0,
+      "seconds": 0.1596787499729544,
+      "side": "changed"
+    },
+    {
+      "pair": 5,
+      "returncode": 0,
+      "seconds": 0.1580816670320928,
+      "side": "changed"
+    },
+    {
+      "pair": 5,
+      "returncode": 0,
+      "seconds": 0.2403820410836488,
+      "side": "baseline"
+    },
+    {
+      "pair": 6,
+      "returncode": 0,
+      "seconds": 0.23771945806220174,
+      "side": "baseline"
+    },
+    {
+      "pair": 6,
+      "returncode": 0,
+      "seconds": 0.15462500019930303,
+      "side": "changed"
+    }
+  ],
+  "schema_version": 1,
+  "shared_preflight_sha256": "63026ec3805f2a3236cc071fc78cf1c54bc6528c43e2f3d82548dc68bf346bce",
+  "shared_source_reader_sha256": "84a45190a6a96c20993a0d4a1045fbf24de660d2e6fef8f8a447bb3d1aa78db6",
+  "summary": {
+    "baseline": {
+      "maximum_seconds": 0.24606208316981792,
+      "median_seconds": 0.2425744580104947,
+      "minimum_seconds": 0.23771945806220174
+    },
+    "changed": {
+      "maximum_seconds": 0.20432179211638868,
+      "median_seconds": 0.1596787499729544,
+      "minimum_seconds": 0.15462500019930303
+    }
+  },
+  "untimed_diagnostic_profiles": {
+    "baseline": {
+      "check_research_commit.py:check": {
+        "calls": 1,
+        "cumulative_seconds": 0.22709825
+      },
+      "check_research_commit.py:history_git": {
+        "calls": 13,
+        "cumulative_seconds": 0.14736337500000002
+      },
+      "check_research_commit.py:main": {
+        "calls": 1,
+        "cumulative_seconds": 0.227757958
+      },
+      "check_research_commit.py:source_history": {
+        "calls": 1,
+        "cumulative_seconds": 0.148056417
+      },
+      "subprocess.py:run": {
+        "calls": 16,
+        "cumulative_seconds": 0.20623616700000003
+      }
+    },
+    "changed": {
+      "check_research_commit.py:check": {
+        "calls": 1,
+        "cumulative_seconds": 0.07309779200000001
+      },
+      "preflight.py:check": {
+        "calls": 1,
+        "cumulative_seconds": 0.12573045800000002
+      },
+      "preflight.py:history_git": {
+        "calls": 5,
+        "cumulative_seconds": 0.052028084
+      },
+      "preflight.py:main": {
+        "calls": 1,
+        "cumulative_seconds": 0.12632416700000001
+      },
+      "preflight.py:source_history": {
+        "calls": 1,
+        "cumulative_seconds": 0.052600209
+      },
+      "subprocess.py:run": {
+        "calls": 8,
+        "cumulative_seconds": 0.106398667
+      }
+    }
+  }
+}

research/joint-source-ancestry/README.md

@@ -0,0 +1,147 @@
+# Joint publication source ancestry
+
+The complete publication byte/history preflight takes **34.2% less time** on the
+unchanged `research/e8-d0-root-swaps` fixture: median **0.2425745 s to
+0.1596787 s**, with the same complete receipt. This is a verifier-engineering
+improvement with a scoped equivalence proof and empirical timing certificate.
+The command is opt-in; it shares the original byte/mode checker and supplies a
+new source-history implementation under this package.
+It changes no mathematical research result, supplier or kappa.
+
+The fixture has 14 files and five distinct source commits. Seven alternating
+fresh-process pairs all pass. Baseline times range from 0.2377195 to 0.2460621 s;
+changed times range from 0.1546250 to 0.2043218 s. The generated
+[BENCHMARK.json](BENCHMARK.json) binds both implementation hashes, the complete
+receipt, every sample and the environment. Timings include interpreter startup,
+argument parsing, file inventory, byte comparison and source-history checking.
+They exclude fixture preparation and separate diagnostic profiling. This is
+warm local filesystem evidence on a shared macOS arm64 host, not a universal
+runtime guarantee or a multiplier speedup.
+
+## What changes
+
+The existing `--check-source-history` preflight resolves every distinct pin and
+runs a separate `merge-base --is-ancestor` query for it. For multiple source
+commits, the changed preflight first checks every exact object type with one
+`cat-file --batch-check` request. It then evaluates:
+
+```text
+git rev-list --max-count=1 <all distinct source commits> --not <publication>
+```
+
+If the result is empty, all pins are ancestors. Otherwise it runs the original
+per-pin checks for the existing first-failure diagnostic. Ordinary missing or
+noncommit objects also take that original path, so a later invalid object does
+not hide an earlier ancestry error. Malformed protocol responses, Git errors
+and timeouts fail closed. The single-source path remains the original path.
+
+On this fixture, source-history Git calls fall from 13 to 5. Separate profiles
+in the generated evidence identify subprocess waiting as the dominant baseline
+cost. The source-history portion occupies about 65% of the profiled complete
+check; removing eight processes is consistent with the measured end-to-end gain.
+The profiles are diagnostic observations, not timed samples.
+
+## Scoped equivalence proof
+
+Let `A(c)` include commit `c` and all its ancestors in the stored Git graph.
+For exact commit pins `p_1,...,p_k` and publication `h`, the revision expression
+above enumerates
+
+```text
+(A(p_1) union ... union A(p_k)) minus A(h).
+```
+
+This set is empty exactly when every `p_i` belongs to `A(h)`. In one direction,
+each pin belongs to its own ancestor set. In the other direction, ancestry is
+transitive, so every ancestor of a pin that belongs to `A(h)` also belongs to
+`A(h)`. Limiting enumeration to one commit preserves emptiness. The type batch
+requires the returned object ID to equal the literal pin and its type to be
+`commit`; an annotated tag cannot silently stand in for its peeled commit.
+This equivalence assumes a stable graph and completed, successful Git queries;
+resource failures still reject.
+
+Only that positive result bypasses the old loop. Ordinary negative answers run
+the unchanged loop in sorted order. Both paths use committed metadata, reject
+shallow history, disable replacement objects and grafts, and retain the existing
+Git environment isolation and 30-second per-command timeout. No new ref, index
+or repository mutation is involved. Corrupt Git graphs and hostile Git binaries
+are outside the original and changed contracts.
+
+The written argument concerns acceptance in the stated finite Git graph; the
+tests exercise concrete positive and failure cases. It is not a formal-kernel
+proof or a re-verification of the fixture's mathematics.
+
+## Verification and reproduction
+
+Run the optimized command from the repository root, using CPython 3.12:
+
+```sh
+/Users/nick/git/integer-mult-bounds/.venv/bin/python3.12 -B research/joint-source-ancestry/preflight.py --package research/e8-d0-root-swaps --check-source-history
+```
+
+Verify and publish this package with:
+
+```sh
+/Users/nick/git/integer-mult-bounds/.venv/bin/python3.12 -B scripts/research_package.py --package research/joint-source-ancestry
+/Users/nick/git/integer-mult-bounds/.venv/bin/python3.12 -B research/joint-source-ancestry/verify.py
+/Users/nick/git/integer-mult-bounds/.venv/bin/python3.12 -B scripts/check_research_commit.py --package research/joint-source-ancestry --check-source-history
+```
+
+The registered normal and optimized commands compare the full real receipt and
+run all 18 source-history tests through the adapter. The thirteen inherited
+contract tests remain the original source; five new tests cover the joint path. They cover merged incomparable sources, duplicate
+pins, foreign cached commits, missing commits, tags/trees/blobs, the first-error
+ordering, committed metadata versus working drift, shallow history, replacements,
+grafts, bounds, malformed responses, Git failures, timeout propagation and the
+read-only CLI. They perform no scientific replay of the benchmark fixture.
+
+To repeat the optional timing experiment, create a scratch directory outside the
+checkout and pass its path:
+
+```sh
+/Users/nick/git/integer-mult-bounds/.venv/bin/python3.12 -B research/joint-source-ancestry/benchmark.py --scratch /absolute/external/scratch
+```
+
+The benchmark authenticates the baseline script against its provenance pin,
+replays that exact module with its original import/root context, and compares
+complete receipts from both CLIs. It needs full Git history and the materialized
+14-file fixture. It writes temporary files only under the supplied directory and
+removes them. The registered verifier does not enforce host-dependent timings.
+The generic replay and publication preflight themselves require Git; this package
+does not promise a Git-free export. No Lean, Docker, selected cold supplier replay
+or complete multiplication theorem was run or claimed.
+
+## Resource bounds and inspection
+
+The existing limits remain 100,000 metadata rows, 256 distinct source commits
+and 16 MiB of source metadata. New loops traverse at most 256 pins and four
+object kinds. A batch request contains at most 16,640 bytes for 64-digit IDs;
+normal replies contain at most 18,688 bytes. Joint reachability emits at most one
+commit ID, with a 30-second timeout. Rejection can add two Git commands before
+the original loop and can therefore be slower. A large-history joint walk might
+also be slower or time out when individual commit-graph checks would pass.
+
+New functions remain under 60 physical lines and use structured, nonrecursive
+control flow. Python allocation is a managed-runtime deviation from fixed-memory
+operation; pin counts, request/reply sizes and benchmark sample counts bound the
+new retained collections. The benchmark's dynamic execution is limited to the
+exact hash-authenticated baseline module; it preserves that module's CLI context.
+Normal and optimized tests run with warnings treated as errors.
+
+## Open next questions
+
+Does shared traversal still help on real publications with 32 to 256 distinct
+source commits and much larger histories? Alternate the complete CLI paths on
+such a publication, keep exact receipts and rejection controls, and profile the
+remaining work. A representative publication and a separate measurement budget
+are needed. No broad performance claim follows from the five-pin fixture.
+
+## Credit
+
+This contribution was produced by Nick Petalas. The publication preflight,
+source-history policy and immutable Git reader retain their existing attribution,
+including ywz65, Nick Petalas, roger-sato, Arham Tahir and Douglas Colkitt.
+Conorholds's root-swap package and all its predecessor mathematics remain their
+authors' work; only its existing file/history inventory is used as a timing
+fixture. New work is Apache-2.0. Independent human and formal proof-kernel review
+are unclaimed.

research/joint-source-ancestry/SOURCES.json

@@ -0,0 +1,36 @@
+{
+  "schema_version": 1,
+  "base_commit": "721c03822c421b7f7fcbb2598993b947c4d88419",
+  "files": [
+    {
+      "path": "certificates/selected-result.json",
+      "usage": "provenance",
+      "git_blob": "d59b5035e7b4cbaa9d688a1d32dd0d6eb79c466a",
+      "sha256": "9b8d0eea95b3bf0dfcb7938a732a6439b87a2685dec2a5ab9e525b3b34debbfa"
+    },
+    {
+      "path": "docs/research/agent-task-catalog.json",
+      "usage": "provenance",
+      "git_blob": "3404fbdd5e7529a0674eef26114b5071e4586cec",
+      "sha256": "ead6cdc9d0af6d58e8a49744d8ec8c7c15e13d2ba93c276c434bcf9b51b93a81"
+    },
+    {
+      "path": "scripts/check_research_commit.py",
+      "usage": "provenance",
+      "git_blob": "dc0be6c743f79b3131e674d92c81b75b069812c9",
+      "sha256": "63026ec3805f2a3236cc071fc78cf1c54bc6528c43e2f3d82548dc68bf346bce"
+    },
+    {
+      "path": "scripts/research_package.py",
+      "usage": "provenance",
+      "git_blob": "48fe9feb4699c09e95505d95c327ea998ac8b3a7",
+      "sha256": "84a45190a6a96c20993a0d4a1045fbf24de660d2e6fef8f8a447bb3d1aa78db6"
+    },
+    {
+      "path": "tests/test_research_source_history.py",
+      "usage": "provenance",
+      "git_blob": "aaaa47c607ce1ee317f0c281586016920ff40eac",
+      "sha256": "0175206fa52dff514fea4d625a072bb66735eb5214be9396dc42b874f0918454"
+    }
+  ]
+}