Skip to content

fix(ci): published figures follow the corpus - #5548

Merged
dmitrii-f-t27 merged 1 commit into
masterfrom
fix/published-figures-repin
Oct 2, 2026
Merged

dmitrii-f-t27 merged 1 commit into
masterfrom
fix/published-figures-repin

Conversation

@dmitrii-f-t27

Copy link
Copy Markdown
Collaborator

What

Published figures -- every pinned spec population re-derived (Spec Guards) reports 9 of 10 figures drifted. It runs after the seal-currency step, so CI has not shown it.

Each figure pins a published number together with the regex that produced it. That regex has not changed once in this file's history (checked commit by commit), so the corpus moved and the question did not.

No pin has moved since d3224e6 (2026-09-16), and five had already drifted at that commit: the merge that resolved this file's conflict kept the older numbers. Since then specs/ grew from 946 to 1146 files. 200 of them are under specs/port/, which did not exist at d3224e6; the port waves added specs without moving these pins.

figure pin at d3224e6 now (769f325) rest of specs/ specs/port/
cast_i8 uses 1079 1079 1081 1079 -> 1079 +2
[]T{} empty slice literals 478 481 525 481 -> 505 +20
x.len() call sites 1319 1319 1414 1319 -> 1301 +113
x.len field reads 680 1075 2197 1075 -> 1567 +630
len(x) call sites 296 296 339 296 -> 332 +7
a::b::c paths 473 474 619 474 -> 503 +116
pub const OP_* 11 61 61 61 -> 61 +0
abs( uses 389 389 418 389 -> 418 +0
test blocks 12783 13056 14350 13056 -> 13419 +931

Each pin now equals the value at 769f325, and each note carries the row above. A comment above test blocks explains the re-pin, matching how earlier moves were explained. --check exits 0 and --self-check passes.

The published numbers themselves (#3489, #3495, #3497, #3501, #3473) are unchanged. The notes say where each one was published and what it is today.

If master moves again before this merges, these numbers move with it; that is what the gate is for.

Refs #5497

🤖 Generated with Claude Code

published_figures.py --check reported 9 of 10 figures drifted on master
769f325. No matcher changed in this file's history, so the corpus moved, not
the question.

No pin had moved since d3224e6 (2026-09-16), and five had already drifted
at that commit: the merge that resolved this file's conflict kept the older
numbers. Since then specs/ grew from 946 to 1146 files, 200 of them under
specs/port/, which did not exist at d3224e6; the port waves added specs
without moving these pins.

Each pin now equals the value at 769f325, and each note gives the value at
d3224e6 and the split of the change between specs/port/ and the rest of
specs/. --check and --self-check pass.

Refs #5497

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@github-actions

github-actions Bot commented Oct 2, 2026

Copy link
Copy Markdown
Contributor

📓 NotebookLM Notebook linked to this PR

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

@github-actions

github-actions Bot commented Oct 2, 2026

Copy link
Copy Markdown
Contributor

PR Dashboard

Generated at: 2026-10-02 10:17:08 UTC

Summary

Status Count
Total Open PRs 50
PRs with Failing Checks 45
PRs with All Checks Green 5
READY 3
FAILING 45
PENDING 0
NO CHECKS YET 0

These columns do not partition: 3 + 45 + 0 + 0 = 48, and there are 50 open PRs. A PR is being counted twice or not at all.

Seal Status

  • ⚠️ STALE -- sha256(compiler.rs)=b7d5cc5c4cf1 != 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).

@dmitrii-f-t27
dmitrii-f-t27 merged commit 4a7fc3b into master Oct 2, 2026
31 of 35 checks passed
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