Skip to content

Six fixed artefact names in one shared TMPDIR: two concurrent corpus runs corrupt each other's columns, with every unres - #4733

Merged
gHashTag merged 2 commits into
masterfrom
queen-3034
Sep 24, 2026
Merged

gHashTag merged 2 commits into
masterfrom
queen-3034

Conversation

@gHashTag

Copy link
Copy Markdown
Owner

Closes #3034

Written by a bee on queen-3034 and published by tools/queen/publish.py. The branch itself is the bee's; the second commit is the coordination entry every pull request must add, which a bee has no way to know about.

2 files changed, 30 insertions(+), 11 deletions(-)

🤖 Generated with Claude Code

Trinity Bee and others added 2 commits September 24, 2026 03:25
The t27c corpus command writes artefacts with fixed names under a shared
/t27-corpus directory, causing race conditions when multiple corpus
runs execute concurrently. This affects measurement accuracy and disarms
the repository's acceptance number control mechanism.

This fix applies the same idiom already used in the codebase (t27-synth-{pid}.v)
to all six fixed filenames:
- c.rs -> c-{pid}.rs
- c.rmeta -> c-{pid}.rmeta
- c.c -> c-{pid}.c
- c.zig -> c-{pid}.zig
- c-test.o -> c-test-{pid}.o
- c.v -> c-{pid}.v

Additionally, the four tests in corpus_unresolved.rs that use the default
TMPDIR are updated to use private TMPDIRs to prevent race conditions
within the test suite itself.

Closes #3034
A pull request must add exactly one docs/now entry and a bee has no way
to know that: its brief names a boundary file and acceptance criteria,
and docs/now/ is neither. The publisher adds it rather than failing the
gate.

Closes #3034

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
@github-actions

Copy link
Copy Markdown
Contributor

PR Dashboard

Generated at: 2026-09-24 03:31:51 UTC

Summary

Status Count
Total Open PRs 40
PRs with Failing Checks 31
PRs with All Checks Green 9
READY 3
FAILING 31
PENDING 0
NO CHECKS YET 0

These columns do not partition: 3 + 31 + 0 + 0 = 34, and there are 40 open PRs. A PR is being counted twice or not at all.

Seal Status

  • ⚠️ STALE -- sha256(compiler.rs)=403499176a5d != manifest seal=87e5cbd3ad94.
    The committed NMSE numbers were certified against an older compiler.rs.
    Run scripts/reseal-check.sh locally for the two-step reseal command (advisory; not a merge gate).

@github-actions

Copy link
Copy Markdown
Contributor

📓 NotebookLM Notebook linked to this PR

This notebook contains session context, decisions, and artifacts for this work.

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.

Six fixed artefact names in one shared TMPDIR: two concurrent corpus runs corrupt each other's columns, with every unresolved counter reading 0

2 participants