fix(gonol): bind complete replay and qualify exact language inputs

Author: erinepshovel-code

.github/workflows/english-gonol-full.yml

@@ -10,6 +10,7 @@ on:
       - "research/english-gonol/tests/test_full_construct_run.py"
       - "research/english-gonol/tests/test_source_order.py"
       - ".github/workflows/english-gonol-full.yml"
+      - "requirements-language-tests.txt"
 
 permissions:
   contents: read
@@ -43,10 +44,10 @@ jobs:
       - name: Set up Python
         uses: actions/setup-python@v7.0.0
         with:
-          python-version: "3.12"
+          python-version: "3.12.14"
 
       - name: Install construction prerequisites
-        run: python -m pip install --disable-pip-version-check pytest PyYAML
+        run: python -m pip install --disable-pip-version-check --require-hashes -r requirements-language-tests.txt
 
       - name: Preflight resources and exact producer identities
         shell: bash

.github/workflows/english-gonol.yml

@@ -6,24 +6,48 @@ on:
       - "research/english-gonol/**"
       - "libs/ucns/**"
       - ".github/workflows/english-gonol.yml"
+      - "tools/qualify_language_gonols.py"
+      - "tools/tests/test_language_qualification.py"
+      - "requirements-language-tests.txt"
   pull_request:
     paths:
       - "research/english-gonol/**"
       - "libs/ucns/**"
       - ".github/workflows/english-gonol.yml"
+      - "tools/qualify_language_gonols.py"
+      - "tools/tests/test_language_qualification.py"
+      - "requirements-language-tests.txt"
+
+permissions:
+  contents: read
 
 jobs:
   english-gonol-tests:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v4
-      - uses: actions/setup-python@v5
+      - uses: actions/checkout@v7.0.1
+        with:
+          persist-credentials: false
+      - uses: actions/setup-python@v7.0.0
         with:
-          python-version: "3.12"
+          python-version: "3.12.14"
+
+      - name: Checkout exact historical lexical test authority
+        uses: actions/checkout@v7.0.1
+        with:
+          repository: The-Interdependency/ucns
+          ref: d7c6f51304ed6c32d48badf63132bea6de8af497
+          path: .ci/ucns-lexical
+          persist-credentials: false
 
       - name: Install pytest
-        run: python -m pip install pytest
+        run: python -m pip install --require-hashes -r requirements-language-tests.txt
 
       - name: English Gonol construction tests
         working-directory: research/english-gonol
-        run: python -m pytest -q tests
+        env:
+          ENGLISH_GONOL_UCNS_ROOT: ${{ github.workspace }}/.ci/ucns-lexical
+        run: python -m pytest -q tests --junitxml="$RUNNER_TEMP/english-gonol.xml"
+
+      - name: Require every test to pass
+        run: python tools/qualify_language_gonols.py check-tests "$RUNNER_TEMP/english-gonol.xml"

.github/workflows/python-gonol.yml

@@ -10,6 +10,9 @@ on:
       - "STACK_MANIFEST.md"
       - "stack-manifest.json"
       - ".github/workflows/python-gonol.yml"
+      - "tools/qualify_language_gonols.py"
+      - "tools/tests/test_language_qualification.py"
+      - "requirements-language-tests.txt"
   push:
     branches: [main]
     paths:
@@ -20,6 +23,9 @@ on:
       - "STACK_MANIFEST.md"
       - "stack-manifest.json"
       - ".github/workflows/python-gonol.yml"
+      - "tools/qualify_language_gonols.py"
+      - "tools/tests/test_language_qualification.py"
+      - "requirements-language-tests.txt"
 
 permissions:
   contents: read
@@ -28,18 +34,19 @@ jobs:
   contracts:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v6
+      - uses: actions/checkout@v7.0.1
         with:
           fetch-depth: 0
-      - uses: actions/setup-python@v6
+          persist-credentials: false
+      - uses: actions/setup-python@v7.0.0
         with:
-          python-version: "3.12"
+          python-version: "3.12.14"
       - name: Install pytest
-        run: python -m pip install pytest
+        run: python -m pip install --require-hashes -r requirements-language-tests.txt
       - name: Stack structural consistency
         run: python tools/check_stack_consistency.py
       - name: Checkout pinned UCNS Public Gonol authority
-        uses: actions/checkout@v6
+        uses: actions/checkout@v7.0.1
         with:
           repository: The-Interdependency/ucns
           ref: 62e08ee1cf3b5d7b6e48c927b1047509e6328b5c
@@ -49,4 +56,6 @@ jobs:
         working-directory: research/python-gonol
         env:
           UCNS_SOURCE_ROOT: ${{ github.workspace }}/.ci/ucns
-        run: python -m pytest -q tests
+        run: python -m pytest -q tests ../../tools/tests/test_language_qualification.py --junitxml="$RUNNER_TEMP/python-gonol.xml"
+      - name: Require every test to pass
+        run: python tools/qualify_language_gonols.py check-tests "$RUNNER_TEMP/python-gonol.xml"

docs/language-gonol-qualification.md

@@ -0,0 +1,68 @@
+# Language construction qualification
+
+Status: execution pending. This protocol qualifies the current English v2
+corpus construction and Python source/control construction. Geometry,
+language understanding, runtime equivalence, and canon promotion remain hmmm.
+
+## Frozen scope and evidence
+
+- Stack starting source: `bcaba85900716e5209150a504660fa2ca5158299`.
+- Minimum repair: `14b6063` (tabs, physical source addresses, density rendering).
+- English: every file admitted by the existing OEWN 2025 loader, from
+  `globalwordnet/english-wordnet@dc343f2683279ecbb13fab4e2fd778d7b162d287`.
+- English UCNS authority: `4f863ad37096b7baab8f62820ad5cb937b62a3a7`.
+- Historical lexical test authority: `d7c6f51304ed6c32d48badf63132bea6de8af497`.
+- Python UCNS authority: `62e08ee1cf3b5d7b6e48c927b1047509e6328b5c`.
+- Python corpus: every tracked `.py` file beneath CPython `Lib/test/` at the
+  `v3.12.14` commit `2abcf904b8dac8c999d2b3aac76681abb333798a`. Exact source inventory is frozen before the run.
+  Files that violate the existing encoding admission rule remain explicit
+  rejection evidence with original byte hashes. Invalid syntax and off-carrier
+  Unicode remain represented, with their existing witness flags and hmmm.
+
+## Acceptance
+
+1. Run every existing English and Python workspace test with exact dependencies;
+   require actual call outcomes, zero skips, zero failures, and nonempty suites.
+2. Rebuild all admitted English input twice in separate processes and empty
+   output directories. Verify SQLite integrity, logical replay, complete counts,
+   unchanged source inventories, and identical logical receipts/manifests.
+3. Enumerate the complete Python corpus before construction. Each eligible file
+   must have two independent process builds, exact source recovery, replay,
+   matching canonical database content and manifest, and an explicit outcome.
+4. Check tab widths and physical addresses against independent source-based
+   witnesses. Verify provenance and constitutive control data, and reject
+   targeted changes to identity, order, source, relation and receipt fields.
+5. Generated databases remain local execution state. Commit compact provenance,
+   outcome and digest receipts; do not substitute a sample for the declared scope.
+
+## Resource preflight and stopping
+
+Observed available resources before execution: 26 GiB free disk, 21.5 GiB RAM,
+Python 3.12.14. The recorded English run needs about 293 MiB output and 280 MiB
+RSS; two serial builds fit. Python file pairs are verified and removed one pair
+at a time, keeping compact per-file evidence. Preflight the largest input and
+storage estimate before starting. These are local persistent processes; no
+arbitrary scientific wall-clock cutoff is imposed. Stop on completion or a
+recorded computational/admission failure; retain failures and all hmmm.
+
+## Usage and next steps
+
+Run the workspace tests from their own directories. Supply
+`ENGLISH_GONOL_UCNS_ROOT` for the exact historical lexical producer and
+`UCNS_SOURCE_ROOT` for the Python producer. The full construction/replay runner
+and final receipt will be linked here after implementation and execution.
+
+Passing this protocol qualifies its explicit construction profiles only.
+UCNS retains geometry authority; Stack retains active construction ownership;
+EDCM retains measurement authority. Independent repository graduation is outside
+this qualification.
+
+## CPython provenance correction
+
+The former `2d7d957248038635334b62b27f650220213d8e5d` pin cannot be fetched
+from the declared upstream (`not our ref`). The annotated `v3.12.14` tag
+`4b65faa0b452113bf130086fb1519be6d9cd9b06` resolves to
+`2abcf904b8dac8c999d2b3aac76681abb333798a`. BASE and the Python work graph
+now use that observed source identity; the graph digest is recomputed.
+No language profile or geometry authority changes. The root Stack manifest
+contains no CPython pin and its authority projection remains applicable.

requirements-language-tests.txt

@@ -0,0 +1,7 @@
+# Exact Linux/Python 3.12 qualification dependencies. Regenerate from downloaded wheels.
+iniconfig==2.3.0 --hash=sha256:f631c04d2c48c52b84d0d0549c99ff3859c98df65b3101406327ecc7d53fbf12
+packaging==26.3 --hash=sha256:d7193f7c8e4e93f444fde0262bf90af30e16fa0ad0ad44cb553c87339b23cd1c
+pluggy==1.6.0 --hash=sha256:e920276dd6813095e9377c0bc5566d94c932c33b27a3e3945d8389c374dd4746
+pygments==2.21.0 --hash=sha256:2363c69b61c4a97c838da3b130dcd6468f4848992b21a82f2a63ec34377137d9
+pytest==8.4.2 --hash=sha256:872f880de3fc3a5bdc88a11b39c9710c3497a547cfa9320bc3c5e62fbf272e79
+pyyaml==6.0.3 --hash=sha256:ba1cc08a7ccde2d2ec775841541641e4548226580ab850948cbfda66a1befcdc

research/python-gonol/BASE.json

@@ -8,7 +8,7 @@
   },
   "project": "python-gonol",
   "python_language_profile": {
-    "commit": "2d7d957248038635334b62b27f650220213d8e5d",
+    "commit": "2abcf904b8dac8c999d2b3aac76681abb333798a",
     "profile": "Python 3.12 file input",
     "repository": "python/cpython",
     "tag": "v3.12.14"

research/python-gonol/README.md

@@ -45,7 +45,7 @@ position.
 One compact SQLite construct plus a small manifest:
 
 ```text
-construct.db   characters, occurrences, control_identities, controls, newlines, meta
+construct.db   characters, occurrences, control_identities, controls, newlines, source_bytes, meta
 manifest.json  schema, source digests, counts, carrier pin, receipt_sha256, hmmm
 ```
 
@@ -76,12 +76,15 @@ verify_construct(Path("construct"), "/home/wayseer_interdependentway_org/src/ucn
 
 ## Acceptance gates
 
-Version 1.1.0 corrects tab expansion and CR-only physical addresses. Rebuild
-1.0.0 constructs from their original sources; replay rejects the old version.
+Version 2.0.0 corrects tab expansion and CR-only physical addresses. Rebuild
+1.x constructs from their original sources; replay rejects the old version.
 Occurrence columns and `controls.start_column` are one-based source-scalar
 addresses. TAB widths use a separate zero-based expanded column, advancing
 through earlier tabs. LF, CR, and CRLF reset that expanded column; FF resets
-the indentation column. CRLF retains both source scalars on the preceding
+the indentation column. Version 2 also preserves original source bytes in the
+database and binds every row in a logical digest. Replay reconstructs the
+complete database in an isolated temporary directory and compares the entire
+manifest and logical digest. CRLF retains both source scalars on the preceding
 physical line and advances the next scalar to line + 1, column 1.
 
 - `x=1\n` produces no not-on-pinned-carrier.

research/python-gonol/WORK_GRAPH.json

@@ -38,12 +38,12 @@
     },
     {
       "authority": "Python 3.12.14 language and CPython recognition witness",
-      "commit": "2d7d957248038635334b62b27f650220213d8e5d",
+      "commit": "2abcf904b8dac8c999d2b3aac76681abb333798a",
       "relation": "exact Python language profile source at tag v3.12.14",
       "repository": "python/cpython"
     }
   ],
   "schema": "the-interdependency.stack-manifest",
   "version": "1.0.0",
-  "work_graph_sha256": "1460f0b437d8edc68d91280d4e8fdbef63d68e410857af8ba92897d4681748f2"
+  "work_graph_sha256": "03d4fab2645602c3ba95d6453d302a53b64cde8ece89a4d688db80041dfd1cdc"
 }

research/python-gonol/docs/PYTHON_AFFIXIATION_BOUNDARY.md

@@ -58,14 +58,19 @@ for source-built participants.
 One compact SQLite construct plus a small manifest:
 
 ```text
-construct.db   characters, occurrences, control_identities, controls, newlines, meta
+construct.db   characters, occurrences, control_identities, controls, newlines, source_bytes, meta
 manifest.json  schema, source digests, counts, carrier pin, receipt_sha256, hmmm
 ```
 
 No giant JSON receipt.
 
 ## Replay
 
+Version 2 preserves original source bytes, including encoding markers, and
+reconstructs the complete construct in temporary storage during replay. Every
+table and manifest field must agree with that reconstruction. Version 1
+receipts must be rebuilt from original source.
+
 Replay verifies, at minimum:
 
 1. pinned Public Gonol digest and module digest;