Repository navigation
test: re-record the pinned corpus baselines for 2.4.0 - #1197
Merged
1 commit merged intoNov 10, 2026
Merged
1 commit merged into
1 commit merged into
Conversation
The baselines were recorded at cf676b8 (16 September). Since then the YAML, JSON, HTML and Liquid indexers, the Ruby and Java test finding, and the C# and nested-scope identity changes moved every repository's numbers on purpose, and fastapi, newtonsoft-json and sinatra fell outside their bands. Recorded with `python -m tests.real_repo_corpus --record`. build_seconds keeps its old values: wall clock depends on the machine, and this one is faster than the one that recorded them, so lowering it would only tighten the time budget by accident. Bands are unchanged. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
code-review-graph reviewOverall risk: 0.00 (LOW) — 0 changed function(s)/class(es), 0 affected flow(s), 0 test gap(s) Token savings: this graph-backed report used ~44,837 fewer tokens (~100%) than reading every changed file in full (estimated, chars/4 approximation). Powered by code-review-graph — local-first analysis; no code leaves the CI runner. |
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Why
The release gate's corpus check compares eight pinned repositories with numbers recorded at cf676b8 (16 September). On
stagingf56618d it reports 3 of 8 repositories out of band (fastapi nodes +72%, sinatra nodes +167% and edges +69%, newtonsoft-json resolved share 0.60 → 0.53), and 8 canary failures follow from that: withfiles_without_nodesnow 0 everywhere, adding 1 no longer goes over the old ceiling. All of these moves are on purpose, so this re-records the baselines for 2.4.0.The newtonsoft-json drop is accepted: calls that cannot be told apart, such as calls to C# overloads, now stay unresolved instead of being bound to a guess.
What changed
tests/corpus_baselines.jsonre-recorded withpython -m tests.real_repo_corpus --record. Same pinned commits, same bands.build_secondskeeps its old values. Wall clock depends on the machine, and this Mac builds 2 to 3 times faster than whatever recorded them, so taking its timings would tighten the time budget by accident (newtonsoft-json's ceiling would fall from 61.7s to 26.2s).Every number that moved, and the PR that moved it
I measured the corpus at cf676b8, at every merge into
stagingsince then, and at every merge inside the two trains that moved anything (#1146 and #1186): 142 builds in all. #1027 is the only other merge that moved anything. The PR contributions in each row add up to the new value. The build at cf676b8 matches the old baseline exactly, and the last build matches the new one.PRs: #1027 Go/Ruby imports, #1092 generic YAML, #1093 workspace imports, #1133
sys.modulesshims, #972 HOC components, #1079 PowerShell, #933 Ruby compact namespaces, #1033 TS abstract classes, #1065 Python@property, #1158.cjs/.kts, #1159 HTML, #878 Liquid, #964 C# namespaces and lexical binding, #1168 nested scope chain, #1151 import cycles (stdlib imports no longer answered by a package file), #1154 Java and Ruby tests intests_for, #715 same-file nodes kept on a qualified-name collision (it only builds with the train fix e7712d7), #1171 honest confidence, #1185 inherited calls, #1174 JSON, #1176 Laravel tests.All moves are improvements (new files, nodes and languages indexed, every parsed file now has nodes, dangling CONTAINS gone, most shares up) except the falls in resolved share: newtonsoft-json −0.072 (#715 −0.053, #964 −0.021) and gson −0.046 from #715, both accepted as above. Also fastapi
imports_resolved_share−0.007 from #1151, where stdlib imports stopped pointing at package files.Proof
promotion_gate.py run --check corpus -- -m corpusgivesfailed (13 passed, 11 failed).passed (24 passed, 0 failed).tests/test_real_repo_corpus_guard.py: 35 passed.ruff check code_review_graph/clean,mypy code_review_graph/no issues.🤖 Generated with Claude Code