Repository navigation
Fix(tests,#1650): bump EXPECTED_PAIR_COUNT 13->14 -- paire ANALYSE-07_en (PR #19943) sans enregistrement - #21383
Merged
Merged
Conversation
…_en arrivee sans enregistrement (PR #19943) La 14e paire detectee par discover_pairs() est SymbolicAI/Lean/ANALYSE/ANALYSE-07-Percolation-Critique_en, mergee le 2026-10-19 (c91ad98) sans mise a jour de la declaration -- le test test_full_repo_state_passes_parity est rouge sur main pur (found 14, declared 13, diagnostic po-2027 dashboard 2026-10-19T06:10Z) et gele toute PR path-filtree scripts/**. Bump + entree d'historique au patron des precedents (#19996/#20392). Validation locale : 35 passed (pytest scripts/translation/tests/). Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
github-actions
Bot
added
the
trivial-diff-advisory
Contributor
|
Trivial-diff advisory (#15740, non bloquant). |
Contributor
|
No organ-duplication: no added def/class collides with another series organ API (scripts/audit/organ_api_index.yaml). Detector: |
myia-ai-01
approved these changes
Oct 19, 2026
myia-ai-01
left a comment
myia-ai-01
left a comment
Collaborator
There was a problem hiding this comment.
Approuvé à la tête c543f57086, réparation de main (périmètre CI, ai-01).
- Scope : 1 fichier, 7 insertions et 1 suppression. Le diff enregistre la 14e paire et rien d'autre.
- Cause vérifiée de mon côté :
git log --diff-filter=Aattribue l'ajout deANALYSE-07-Percolation-Critique_en.ipynbà c91ad98 (#19943). Le jobScripts Tests (CPU)de #21365 rendfound 14, declared 13sur le runnermyia-ai-01-wsl-3à 07:35Z. - Second rouge cité au body : il est levé. #21366 est sur
main(901825f), ettest_detect_code_in_markdown_cells.pypasse en local surmain9ed2c39 (13 passed). - Note pour l'instrument : en local, sur un clone qui porte un
.tmp/ignoré,discover_pairscompte 15 paires (unfort-boyard-python_en.ipynbégaré sous.tmp/). La CI ne voit pas ce dossier et compte 14. Ce n'est pas un défaut de la PR, je le note pour quiconque re-mesure en local. - Suites : #21296 et #20711 déclarent aussi un passage 13 -> 14. Après ce merge, chacune devra passer à 15, puis 16, dans l'ordre où elles arrivent.
This was referenced Oct 19, 2026
Closed
Closed
This was referenced Oct 19, 2026
This was referenced Oct 19, 2026
This was referenced Oct 19, 2026
jsboige
added a commit
that referenced
this pull request
Oct 19, 2026
…PAIR_COUNT 14 -> 15 Conflit de registre : les deux lignees ajoutait chacune une entree « 13 + 1 = 14 » pour un 14e paire distinct (branche : research_archive_en T4 ; main : ANALYSE-07_en via #19943, enregistree par #21383). Resolution : les deux entrees conservees + entree de resolution, EXPECTED_PAIR_COUNT = 15 (main 14 + research_archive_en). Preuve : pytest test_check_translation_parity.py 35/35 passed sur l arbre fusionne (test_full_repo_state_passes_parity trouve 15 paires). Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
myia-ai-01
pushed a commit
that referenced
this pull request
Oct 20, 2026
* Add(ci,#21391): EXPECTED_PAIR_COUNT guard at PR time `main` has now turned red for the fourth time because a notebook PR adds a new `*_<lang>.ipynb` pair without bumping the constant `EXPECTED_PAIR_COUNT` in `scripts/translation/tests/test_check_translation_parity.py`. History of the four rouges: #19996, #20392, #20453, #19943 (constant updated in each by a follow-up PR). The test that compares the count (`test_full_repo_state_passes_parity`) lives in `Scripts Tests (CPU)`, filtered on `scripts/**`, so a notebook PR that doesn't touch `scripts/` doesn't even trigger it. `translation-parity.yml` runs only nightly. The invariant "pair added/removed/renamed => constant bumped in the same PR" has no carrier at PR time. This guard is the carrier. Pipeline (pure git, no clone, no kernel): 1. `git diff --name-status -M50% base..head -- 'MyIA.AI.Notebooks/...'` lists notebook A/D/R status (M is ignored). 2. For each A/D/R translation file `*_<lang>.ipynb`, the guard reads the corresponding FR source at head via `git show head:path` (no disk walk). Files that pair are recorded. 3. If the diff adds or destroys at least one pair AND `EXPECTED_PAIR_COUNT` was not modified in the same diff: ROUGE, with the exact line to edit, the new value expected, and the 5-line history-entry pattern already in use by the file. 4. If the diff modified the constant, the new value is compared to `discover_pairs(repo_root, langs)` (the disk walk, only on this branch). ROUGE if the values disagree. Acceptance (re-run on the SHAs cited in the issue): 1. POSITIVE on #19943 (c91ad98^..c91ad98): ROUGE, names `ANALYSE-07-Percolation-Critique_en`, expected bump 13 -> 14. 2. NEGATIVE on #21383 (origin/main..pr-21383): GREEN (no pair diff). 3. M-only diff (existing _en.ipynb modified): GREEN, A/D/R set excludes M by design. 4. .tmp/ orphan on local disk: the guard reads git diff, not disk; the orphan is invisible to it. 5. Cost: 0.35s (no disk count) / 0.35s (with disk count) on this machine; no blob:none clone of 2 GB. 25/25 pytest tests PASSED. Out of scope: the workflow integration lives in `.github/workflows/always-on-guards.yml` -- ai-01 does the merge (claim `paths: scripts/translation/**` only; workflow file excluded). Refs: #21391, #1650, #19943, #21383, NAMING c.12811 (encoding), NAMING c.530 strict (base-inherited), NAMING c.971 strict (0 re-ping). Co-Authored-By: Claude Haiku 4.5 (1M context) <noreply@anthropic.com> * fix(translation,#21391): test_check_pair_count_declared setup_class path Windows hardcode -> dynamique Le setup_class TestAcceptance utilisait Path(r"D:\CoursIA-2") qui est un chemin Windows hardcode. Sur un runner CI Linux, ce chemin n'existe pas (Errno 2 FileNotFoundError PosixPath('D:\CoursIA-2')). Fix : Path(__file__).resolve().parents[3] pour remonter du fichier test au repo root (scripts/translation/tests -> 3 niveaux de parents). NAMING c.91 / c.17071 strict : encage les tokens bloquants. NAMING c.16866 ★★ OK : pre-commit OK (encoding='utf-8' sur subprocess). NAMING c.11064 strict : paths: scripts/translation/** only. Tests : 25/25 PASSED localement (acceptance + integration + helpers). Co-Authored-By: Claude Haiku 4.5 (1M context) <noreply@anthropic.com> * fix(translation,#21391): derive constant line in ROUGE msg, address Hermes F1/F2 F2 -- _rouge_message no longer hardcodes :731 (the constant has already drifted to :737). _constant_line_number_in_head resolves the current line from the head text; on failure it prints the search pattern instead of a stale number. New witness asserts the resolved line == the live line and != the stale 731. F1 -- wiring stays the dispatcher's decision (unchanged); the body reformulation is delivered in a follow-up comment on the PR. Co-Authored-By: Claude Haiku 4.5 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Haiku 4.5 (1M context) <noreply@anthropic.com>
This was referenced Oct 20, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Grain: LIGHT/guard -- lane myia-po-2025:CoursIA -- prev: MED/notebook-python #21382
Repair main — un des deux rouges Scripts Tests (CPU)
Diagnostic (mesuré sur main pur f52f029, corroboré par po-2027 au dashboard 2026-10-19T06:10Z) :
test_check_translation_parity.py::test_full_repo_state_passes_parityéchoue avecfound 14, declared 13. L'organecheck_translation_parity.pyrend « OK : 14 paires, 0 violation » — seul le compte déclaré (EXPECTED_PAIR_COUNT = 13, l.731) est périmé.Cause racine : la 14e paire —
SymbolicAI/Lean/ANALYSE/ANALYSE-07-Percolation-Critique_en— est arrivée sur main le 2026-10-19 (commit c91ad98, PR #19943, pli 3 Origami #19898) sans mise à jour de la déclaration. Même classe que les précédents #19996/#20392 : la paire arrive avec son carnet, la déclaration ne bouge pas dans la même PR.Fix : bump
EXPECTED_PAIR_COUNT = 13→14+ entrée d'historique au patron établi (7 insertions, 1 fichier).Validation
python -m pytest scripts/translation/tests/test_check_translation_parity.py: 35 passed (donttest_full_repo_state_passes_parity).python scripts/translation/check_translation_parity.py: OK, 14 paires, 0 violation.Périmètre et résiduel
detect_code_in_markdown_cells.py, Lean-19c cell#8) est réparé par Fix: Lean-19c formules E1/E3 en LaTeX -- detect_code_in_markdown_cells vert sur main (rouge base-inherited) #21366 (ouverte, CI verte, en attente merge ai-01) : le job Scripts Tests sur CETTE PR restera rouge pour cette jambe base-inherited tant que Fix: Lean-19c formules E1/E3 en LaTeX -- detect_code_in_markdown_cells vert sur main (rouge base-inherited) #21366 n'est pas mergé — imputation documentée ici.See #1650 (famille translation, sans clôture).
🤖 Generated with Claude Code