Skip to content

spec: ci-test-shards - #274

Open
yihanzhu wants to merge 5 commits into
mainfrom
ystack/spec/ci-test-shards
Open

spec: ci-test-shards#274
yihanzhu wants to merge 5 commits into
mainfrom
ystack/spec/ci-test-shards

Conversation

@yihanzhu

Copy link
Copy Markdown
Owner

Tracks #269

G2 spec for the merged intent (G1, PR #270). Frontmatter records intent-blob e56b92429ca56a638827ebf7d0e270ffa2bd1089 and risk: high (the change edits .github/workflows/ci.yml, the merge gate's workflow and a constitution path). Merging this PR accepts both. It does not close the intake issue.

This PR adds only work/ci-test-shards/spec.md.

@cloudflare-workers-and-pages

cloudflare-workers-and-pages Bot commented Sep 10, 2026

Copy link
Copy Markdown

Deploying ystack with  Cloudflare Pages  Cloudflare Pages

Latest commit: 7ec0664
Status: ✅  Deploy successful!
Preview URL: https://4d7c60f8.fabrica-6yx.pages.dev
Branch Preview URL: https://ystack-spec-ci-test-shards.fabrica-6yx.pages.dev

View logs

@yihanzhu

Copy link
Copy Markdown
Owner Author

Codex reviewer (cross-vendor, read-only)

Reviewed-head: 45a92ef
Reviewed-base: 4965175
reviewer: gpt-5.5 @ high

Posted verbatim by codex-review.sh (codex exec review --json --base refs/codex-review/274-98076/base in an isolated temp worktree, sandbox forced read-only). Comments only — Codex never pushes, approves, or merges.

The new spec is hash-linked to the accepted intent, records high risk, and stays within the artifact path. I did not find a discrete blocking issue in the proposed requirements or process.

@yihanzhu

Copy link
Copy Markdown
Owner Author

Codex reviewer (cross-vendor, read-only)

Reviewed-head: 1ba4bde
Reviewed-base: 40582c2
reviewer: gpt-5.5 @ high

Posted verbatim by codex-review.sh (codex exec review --json --base refs/codex-review/274-6458/base in an isolated temp worktree, sandbox forced read-only). Comments only — Codex never pushes, approves, or merges.

The added spec contains an impossible acceptance condition because it also requires adding a new discovered test suite. That contradiction should be fixed before the spec is accepted as the implementation contract.

Review comment:

  • [P2] Remove the impossible output-diff requirement — /private/var/folders/rm/bkx4f4b91vqfn79vk02pcd140000gn/T/tmp.ncmVsKLDHp/work/ci-test-shards/spec.md:27-30
    This verification cannot pass once requirement 9 is implemented: the new executable scripts/test/run-all-sharding.test.sh will be discovered by the unchanged find ... '*.test.sh' rule and run in no-argument mode, so the output and final count will gain at least that suite instead of diffing exactly against today's output. The spec should require preserving the existing suites' behavior/order, or explicitly account for the new focused test.

@yihanzhu

Copy link
Copy Markdown
Owner Author

Codex reviewer (cross-vendor, read-only)

Reviewed-head: 731f7c5
Reviewed-base: 40582c2
reviewer: gpt-5.5 @ high

Posted verbatim by codex-review.sh (codex exec review --json --base refs/codex-review/274-14100/base in an isolated temp worktree, sandbox forced read-only). Comments only — Codex never pushes, approves, or merges.

The spec contains a scope contradiction that would make the required sharding regression test appear out-of-scope during implementation. This should be clarified before accepting the artifact.

Review comment:

  • [P2] Align the out-of-scope list with the required test — /private/var/folders/rm/bkx4f4b91vqfn79vk02pcd140000gn/T/tmp.L4OCwc4vff/work/ci-test-shards/spec.md:175-176
    When this spec becomes the accepted G2 scope, requirement 9 requires adding scripts/test/run-all-sharding.test.sh, but this out-of-scope bullet forbids any change to which suites exist or which CI runs. Because this repo treats spec.md as the hash-linked artifact required before planning (work/README.md:26-29), the implementation is both required and out-of-scope; clarify this to mean no change to existing suites, or explicitly exempt the focused sharding test.

@yihanzhu

Copy link
Copy Markdown
Owner Author

Codex reviewer (cross-vendor, read-only)

Reviewed-head: 7ec0664
Reviewed-base: 40582c2
reviewer: gpt-5.5 @ high

Posted verbatim by codex-review.sh (codex exec review --json --base refs/codex-review/274-39156/base in an isolated temp worktree, sandbox forced read-only). Comments only — Codex never pushes, approves, or merges.

The spec otherwise follows the high-risk artifact shape, but it adds a discovered test suite despite the accepted intent explicitly keeping the suite set and no-argument runner behavior unchanged.

Review comment:

  • [P2] Keep the no-argument suite set in scope — /var/folders/rm/bkx4f4b91vqfn79vk02pcd140000gn/T/tmp.WpmSJWWMEU/work/ci-test-shards/spec.md:34-35
    Under the artifact chain, the spec is the next hash-linked artifact after the accepted intent (AGENTS.md:370-379), and that intent says “No change to which suites exist” and keeps the no-argument runner unchanged (work/ci-test-shards/intent.md:33-37). Requiring a new discovered *.test.sh suite widens that accepted scope and intentionally changes no-selector output; either re-accept the intent for the added suite or keep the sharding check outside the discovered suite set.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant