chore(ci): regenerate the .gitea twins stale since v0.57.0 #1155

Merged

alex merged 1 commit from i/gitea-twins-stale-after-v0570 into main

2026-09-05 01:16:12 +02:00

quartermaster commented

2026-09-05 01:15:29 +02:00

Owner

rt gitea-twin --check fails on 7 twins — and it fails identically on main. A clean checkout of forgejo/main run through the same binary reports the same 7 files, so this is inherited, not introduced by any open PR.

twins   BUILD_BAKED_TOOLKIT_REF: 'v0.57.0'    the cut's value
source  BUILD_BAKED_TOOLKIT_REF: 'main'       reset by 89f9dc8

89f9dc8 chore: post-cut bookkeeping for v0.57.0 reset the .forgejo sources back to main after the cut and did not regenerate the twins. It carried [skip ci], so gitea-twin-check never graded it — main has been red here since, and every PR opened since inherits the failure.

That is the same cut that published v0.57.0 with zero assets (#1154). Two independent defects from one event, both hidden: one by a step-ordering bug, one by [skip ci].

Produced by rt gitea-twin --write, no hand edits. Seven files, one line each.

Verified: --check goes rc=1, 7 out-of-date → rc=0, 0 out-of-date.

Split from #1154 deliberately — that PR fixes the asset-ordering bug and unblocks a broken published release; folding an unrelated seven-file regeneration into it would make the one that needs to be obviously correct harder to read.

🤖 Generated with Claude Code

https://claude.ai/code/session_01LUEggQMJjaizj2nFVofeyH

`rt gitea-twin --check` fails on 7 twins — **and it fails identically on `main`**. A clean checkout of `forgejo/main` run through the same binary reports the same 7 files, so this is inherited, not introduced by any open PR. ``` twins BUILD_BAKED_TOOLKIT_REF: 'v0.57.0' the cut's value source BUILD_BAKED_TOOLKIT_REF: 'main' reset by 89f9dc8 ``` `89f9dc8 chore: post-cut bookkeeping for v0.57.0` reset the `.forgejo` sources back to `main` after the cut and did not regenerate the twins. **It carried `[skip ci]`, so `gitea-twin-check` never graded it** — main has been red here since, and every PR opened since inherits the failure. That is the same cut that published v0.57.0 with zero assets (#1154). Two independent defects from one event, both hidden: one by a step-ordering bug, one by `[skip ci]`. Produced by `rt gitea-twin --write`, no hand edits. Seven files, one line each. **Verified:** `--check` goes `rc=1`, 7 out-of-date → `rc=0`, 0 out-of-date. **Split from #1154 deliberately** — that PR fixes the asset-ordering bug and unblocks a broken published release; folding an unrelated seven-file regeneration into it would make the one that needs to be obviously correct harder to read. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_01LUEggQMJjaizj2nFVofeyH

quartermaster added 1 commit

2026-09-05 01:15:29 +02:00

chore(ci): regenerate the .gitea twins stale since v0.57.0 (#1112 ) fe8f77c3a7

`rt gitea-twin --check` fails on 7 twins, and it fails identically on main — a
clean checkout of forgejo/main reports the same 7 files, so this is inherited
rather than introduced by any open PR.

  twins   BUILD_BAKED_TOOLKIT_REF: 'v0.57.0'   the cut's value
  source  BUILD_BAKED_TOOLKIT_REF: 'main'      reset by 89f9dc8

89f9dc8 "post-cut bookkeeping for v0.57.0" reset the sources and did not
regenerate the twins. It carried [skip ci], so nothing graded it — the same
cut that published v0.57.0 with zero assets also left main red here.

Produced by `rt gitea-twin --write`; no hand edits. Verified: check goes
rc=1/7-out-of-date to rc=0/0.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LUEggQMJjaizj2nFVofeyH