Skip to content

Review backlog (correctness): replaySupersedeOntoBase claim is not keyed by prior_copy and is journal-replayed on re-adoption #15225

Closed

Description

@hivecommons-hive

Blocked on PR #15209. This finding came from reviewing PR #15209 and is about code that PR is changing. Fix it in the PR, or leave it: the hive lifts the block when the PR merges, and unblocks this item for re-triage if the PR is closed without merging. Once it is unblocked, check the finding still applies on the base branch; if it is already fixed, close this item without a PR.

Filed from an out-of-scope finding in the Hive review of PR #15209.

PR: #15209 (#15209)
Perspective: correctness
Severity: low (P3)
Evidence: src/pkg/github/automerge/automerge_contributor_adopt.go:950 (

)
Finding: replaySupersedeOntoBase claim is not keyed by prior_copy and is journal-replayed on re-adoption

The branch_push claim in replaySupersedeOntoBase uses Inputs {pr, head_sha, base, reason} with no prior_copy (lines 950-951), so on a re-adoption after a closed copy the journal replays it as already applied and the base merge into hive/adopt-N silently does not happen; the fresh copy may open with the conflict the first adoption resolved. Pre-existing; the PR's scope covers the push/create claims that caused the revert loop.

Fingerprint: 8cdd42a324d9afb375279e3f4756bd8abcb974eb7478f4011641fa59eb728e98

filed by Hive review backlog

Activity

  1. added
    blockedWaiting on something outside this repository; not contributor work until a human clears the label
    on Oct 29, 2026
  2. added
    help wantedDenotes an issue that needs help from a contributor. Must meet "help wanted" guidelines.
    on Oct 29, 2026
  3. removed
    help wantedDenotes an issue that needs help from a contributor. Must meet "help wanted" guidelines.
    blockedWaiting on something outside this repository; not contributor work until a human clears the label
    on Oct 29, 2026
  4. hivecommons-hive commented on Oct 29, 2026

    @hivecommons-hive
    ContributorAuthor

    Unblocked: #15209 is no longer open. This issue is actionable again.

  5. hivecommons-hive commented on Oct 29, 2026

    @hivecommons-hive
    ContributorAuthor

    Additional finding (folded in — same finding area and category). Raised in the Hive review of PR #15206 (#15206) and folded into this issue because it is a correctness finding at the same symbol or lines of src/pkg/github/automerge/automerge_contributor_adopt.go as this issue. Address it together with this issue; if it turns out to be a different problem, file it separately before closing.

    Review backlog (correctness): Journal replay chain resolves only one level

    • Journal replay chain resolves only one level (src/pkg/github/automerge/automerge_contributor_adopt.go:866)

    Perspective: correctness
    Severity: low (P3)
    Evidence: src/pkg/github/automerge/automerge_contributor_adopt.go:866 (

    )
    Finding: Journal replay chain resolves only one level

    After openCopy(replayed) at :866, a second journal replay (store cleared twice at an unchanged head) yields created==nil, err==nil and falls to the 'adoption copy not opened' error at :872 every pass instead of claiming afresh. Identical in v5's cf1fe3a copy, so this is a follow-up on v5 (bounded loop), not a blocker here.

    Fingerprint: c6a3fd57e7e359b0c979652140ea32350912229ad9825799cf40bb208ee1b8f3

    filed by Hive review backlog

  6. added
    help wantedDenotes an issue that needs help from a contributor. Must meet "help wanted" guidelines.
    on Oct 29, 2026
  7. hivecommons-hive commented on Oct 29, 2026

    @hivecommons-hive
    ContributorAuthor

    Unblocked: #15209 merged. This item is actionable now; check the finding still applies on the base branch first, and close it without a PR if it is already fixed.

  8. hivecommons-hive commented on Oct 29, 2026

    @hivecommons-hive
    ContributorAuthor

    🔒 Claimed by agent scanner (agent on hive hosted-available-oke-11-placeholder-r05x). The claim lasts 2h0m0s without activity and renews while the work continues. Reply /unclaim to release, or /claim to take it over if you outrank the holder.

  9. hivecommons-hive commented on Oct 29, 2026

    @hivecommons-hive
    ContributorAuthor

    Additional finding (folded in — same finding area and category). Raised in the Hive review of PR #15294 (#15294) and folded into this issue because it is a correctness finding at the same symbol or lines of src/pkg/github/automerge/automerge_contributor_adopt.go as this issue. Address it together with this issue; if it turns out to be a different problem, file it separately before closing.

    Review backlog (correctness): Merge-time close in settleSuperseder remains ungated

    • Merge-time close in settleSuperseder remains ungated (src/pkg/github/automerge/automerge_contributor_adopt.go:1324)

    Perspective: correctness
    Severity: info (P3)
    Evidence: src/pkg/github/automerge/automerge_contributor_adopt.go:1324 (

    if _, _, err := c.gh.PullRequests.Edit(ctx, obs.Owner, obs.Repo, obs.Number, &gh.PullRequest{State: gh.Ptr("closed")}); err != nil {
    )
    Finding: Merge-time close in settleSuperseder remains ungated

    A gate-refused close-first falls to the keep-open path, whose merge-time close (PullRequests.Edit closed) is still a baselined raw write, so the original is closed ungated anyway. Acknowledged in the PR body as the automerge_contributor_adopt.go follow-up.

    Fingerprint: 392ccd2e0490394ab3f5b755bcdca453eefacad4676af67d41bf61bbc7ffaab7

    filed by Hive review backlog

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

Metadata

Metadata

Assignees

No one assigned

    Labels

    agent/reviewerFiled by the reviewer agenthelp wantedDenotes an issue that needs help from a contributor. Must meet "help wanted" guidelines.

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions