Repository navigation
Pin the fork's integration branch (ADR-0001) - #66
Conversation
The subject becomes a commit of getwyrd/client-rust integration: upstream 3408d95 plus #557, #551 and #558. pins.toml gains branch and upstream_base; check-pins.sh checks rev is on integration and upstream_base is on both rev and upstream master. CI clones the fork and adds an informational parity run at upstream_base. rust-driver handles #557's MultipleRegionErrors, with a pre-557 feature for the upstream run; wyrd-gate drops ExtractedErrors. Signed-off-by: Eduard Ralph <eduard@ralphovi.net>
Strict make verdict, make ledger and the upstream view at 3408d95, captured on a leased v8.5.5 cluster at harness 848d24d. Signed-off-by: Eduard Ralph <eduard@ralphovi.net>
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: f738fd4b9a
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Plain merge-base can return a carried PR's tip once upstream merges that PR with a merge commit: a side-branch state master never was. Take the newest commit on upstream master's first-parent chain that rev contains instead. Signed-off-by: Eduard Ralph <eduard@ralphovi.net>
Signed-off-by: Eduard Ralph <eduard@ralphovi.net>
|
@codex review |
|
Codex Review: Didn't find any major issues. Another round soon, please! Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. Codex can also answer questions or update the PR. Try commenting "@codex address that feedback". |
Moves the client-rust subject from upstream tikv/client-rust
masterto the fork'sintegrationbranch, so the loop can build on its own fixes before upstream merges them. Decision record: ADR-0001 (status Proposed).Pin
pins.toml [client_rust]:rev = b10bc69(integration: upstreammaster@3408d95plus *: split ExtractedErrors into MultipleKeyErrors and MultipleRegionErrors tikv/client-rust#557, #551, #558 and the fork'sAGENTS.md, AGENTS.md: the client-rust-pdca review rubric (fork-only) client-rust#1), newbranchandupstream_base = 3408d95.check-pins.sh:revmust be on the first-parent chain of<fork>/integration(a state the branch has been at, not a merged-in PR branch's commit), andupstream_basemust be the newest commit on upstreammaster's first-parent chain thatrevcontains. UnderPARITY_STRICTa check that cannot run fails.CI
masterforcheck-pins.sh(the gate job re-running it is CI: the gate job does not re-run the pin's ancestry checks before producing ledger evidence #67).upstream-viewjob: runs the parity diff atupstream_basewithPARITY_STRICT=0,continue-on-error, and uploads it. It never gates and is not ledger evidence (ledger-checkrefuses a non-strict run).rust-driverfeaturepre-557swaps theExtractedErrors/MultipleRegionErrorsmatch arms so the driver compiles against both trees; only the upstream view enables it.wyrd-gatedrops theExtractedErrorsarm.Verdict at b10bc69 (evidence, strict, v8.5.5 lease): gate 18 + 1 passed, as expected; ledger G-0001 HOLDS, G-0002 XDIVERGE as declared; upstream view at
3408d95matches. README verdict restated; the3408d95verdict moves to History.Reviewed before push: batched Codex review (clean) and an adversarial review (its two P1s fixed: the first-parent check and the evidence file; its P2 on the ADR overstating branch protection fixed in the text).
Open question for the maintainer:
integration's protection disables force-pushes but does not bind admins and requires no status check; the ADR now says so.