Repository navigation
java-compile: keep the engine scan across runs (Part of #5142) - #6605
Merged
Merged
Conversation
…che (Part of #5142) java-base-generation.yml seeds a content-keyed scan cache per corpus, java-compile.yml restores it (restore-only) and passes --scan-cache to the unchanged check and the engine-scan build. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
squid-protocol
added
shepherd:merge
Contributor
|
No description provided. |
Contributor
Contributor
|
No description provided. |
Contributor
|
No description provided. |
2 tasks
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Wires
--scan-cache(added in #6375) into the Java workflows so the dsf engine scan (~90 s per side) is restored instead of re-run.java-base-generation.yml: newjava-scan-<corpus>-<hash>cache (lookup-only check,--scan-cacheon the seeding run, main-only save). Generation is re-run when either the base-generation entry or the scan entry is missing.java-compile.yml: key step + restore-only (ci: java-same-base and gradle-home caches save near-duplicate entries (sibling of #4847) #4968),--scan-cacheon the "unchanged?" check and the engine-scan build. Fallbackrestore-keysprefix per corpus; the tool'sscan_keyis the exact check, so a coarse Actions key can only cost a rescan, never a wrong scan.cics_task_containers.pykept in) + corpus pin + matrix tool. PR hashes the base checkout, push its own.tests/tools/test_cache_keys.py: pins restore/save split, equal hash inputs on both sides, flags passed.Cross-repo: none.
Measurement (ci-workflows rule 5): no new job; adds one lookup-only cache restore per seeding run. Expected saving on a warm hit: up to ~90 s per side on dsf in the unchanged check. Not measured on CI yet (first effect after the seeding workflow runs on main). Plan reach unchanged (same path filters).
Checklist: partial [c:7a2199c1] -- the seed fires only on translator/controller-path pushes, so after an engine-only main commit a PR still misses and rescans; the dsf ticket pass (~20 s/config) remains.
Gates: pr_gates --fast 7/7, audit_check clear, workflow tests 391 pass.
Lessons for agents
hashFilesjoined with-inside one${{ }}is subtraction: one${{ }}per part.🤖 Generated with Claude Code
Class / Sibling gaps
Copied from the worker's report by promoter-bot's precheck.
Sibling gaps