You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
spec-guards: published_figures pins drifted since 5b2f8e478 (8 figures) #6899
spec-guards runs python3 tools/published_figures.py --check, and it is red on master: 8 of the 10 pinned spec figures drifted. The pins were last moved at 5b2f8e4 (2026-10-02, #5613), and every figure at that commit still equals its pin. The corpus moved after that and no PR re-pinned.
Cause
The matchers have not changed. Each movement comes from merges since 5b2f8e4, counted per merge with the file's own regexes and the same rule for skipping comments:
The pins in tools/published_figures.py follow the corpus. Each pin gets a provenance note in the file's existing style (old -> new at <sha>: ...).
The matchers, the --check exit codes and --self-check are unchanged.
spec-guards is green.
Boundary
tools/published_figures.py is Python, so editing it needs the owner-approved-foreign label. Under the owner's standing rule of 2026-10-06, the agent adds that label and does the work. The edit changes data only: the pinned integers and their provenance strings, and no logic. It is listed in tools/policy/foreign-exceptions.txt until the PR merges. The fix lands in PR #6894.
spec-guards "Published figures" is red on master again, and PRs inherit it (seen on #7091). Measured with python3 tools/published_figures.py --check at each master commit since the last re-pin, measured up to 2026-10-06T21:05Z:
The re-pin commit itself already drifts: #6969 was measured on its own base, and the Queen merges that landed just before it (#6757, #6815, #6817) moved both figures without re-pinning. Every later merge adds test blocks, and none moves x.len().
The same drift shows on master's own run 37524976710 (96cf7c8) and on #7091's run 37528654807. So this red belongs to master, not to that PR. I have not touched the pin: it belongs to this lane.
Activity
spec-guards: published figures follow the corpus; harness learns enum…
published figures: test blocks 15611 -> 15614 after master merged #6892…
published figures: re-pin after master moved to 33e61b3 (Refs #6899)
published figures: test blocks 15714 -> 15720 after master merged #6965…
fix(numeric): settle gf16.t27 against FORMAT-SPEC-001 -- no subnormal…
feat(t27b): std.mem.eql/indexOf on byte slices and byte-array pointers (
gHashTag commentedon Oct 6, 2026
spec-guards "Published figures" is red on master again, and PRs inherit it (seen on #7091). Measured with
python3 tools/published_figures.py --checkat each master commit since the last re-pin, measured up to 2026-10-06T21:05Z:The re-pin commit itself already drifts: #6969 was measured on its own base, and the Queen merges that landed just before it (#6757, #6815, #6817) moved both figures without re-pinning. Every later merge adds test blocks, and none moves x.len().
The same drift shows on master's own run 37524976710 (96cf7c8) and on #7091's run 37528654807. So this red belongs to master, not to that PR. I have not touched the pin: it belongs to this lane.
gHashTag commentedon Oct 6, 2026
Landed on master via #6969 (2116d20), which carried #6894's commits.
fix(codegen): gen-rust enum order and gen-zig float-call cast (#6941) (…