docs(CHANGELOG): backfill 0.506.0-0.509.0, which shipped unrecorded - #1474
Merged
Conversation
…skip actions] VERSION said 0.509.0 while the newest CHANGELOG heading was [0.505.0]: four releases shipped with no section at all. The cause is structural. The release pipeline renames `## [current]` to the next version number, so a merge that adds no `[current]` section means the release records NOTHING — silently, with no warning. Of the five feature PRs merged in that window only #1464 carried a CHANGELOG entry; #1465, #1467, #1469 and #1472 did not. That is my omission across three of them. Reconstructed from the merge log rather than from memory. Each release window turned out to contain exactly one PR, so attribution is unambiguous: 0.506.0 #1465 nightly install refresh + git fetch --tags 0.507.0 #1467 Windows make install dropped the .exe suffix 0.508.0 #1469 fs.read / fs.read_binary error detail (+ the directory false-success bug found while testing it) 0.509.0 #1472 contrib/avcodec + the contrib-check link wiring Entries are written from the shipped code, not transcribed from commit subjects, and each keeps the detail that makes it useful later: why `ae --version` cannot detect a broken Windows install, why a sandbox refusal needs different wording from an I/O error, why the nightly's stale-toolchain failure reads like a GCC 16 bug and is not. Additions only — 122 lines, no existing section touched. Verified afterwards that VERSION now matches the newest heading and that none of the four new headings duplicate. (The pre-existing duplicate [0.435.0] and [0.497.0] headings are untouched; they are the same pipeline bug showing up earlier and are worth a separate fix so the rename cannot collide with an existing version.) Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This was referenced Aug 9, 2026
paul-hammant
added a commit
that referenced
this pull request
Aug 9, 2026
The avcodec audio-decode commit landed with no CHANGELOG entry and there was no [current] section, so per this file's own workflow the next release would have tagged a version with whole-track audio decode unmentioned — the same way 0.506.0 through 0.509.0 shipped empty (backfilled in #1474, root cause in #1477). Written from the shipped code rather than the commit subject, and pairs the entry with load_pcm (0.512.0) since the two only make sense together: this is the producer, that is the consumer, and asks/pcm-please.md wanted both. Verified end-to-end here before writing it, rather than transcribing the commit's numbers: a 2s AAC clip decodes to 356352 bytes at 44100 Hz stereo and audio.load_pcm reports duration_ms=2020. (The ~1% over 352800 is AAC encoder padding, not a defect.) [skip actions] deliberately NOT used here, despite this being the docs commit: it is the branch HEAD, so the token would skip CI for the whole branch — including 244 lines of new C in contrib/avcodec that wants exercising on the Windows and macOS runners. The token is for branches that are docs-only end to end, not for the docs commit of a branch that ships code. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
VERSIONsaid 0.509.0 while the newest CHANGELOG heading was[0.505.0]— four releases shipped with no section at all.Why it happened
Structural, not an oversight in one PR. The release pipeline renames
## [current]to the next version number, so a merge that adds no[current]section means the release records nothing — silently, with no warning.Of the five feature PRs merged in that window, only #1464 carried a CHANGELOG entry. #1465, #1467, #1469 and #1472 did not. Three of those four are mine.
Attribution
Reconstructed from the merge log rather than memory. Each release window contained exactly one PR, so the mapping is unambiguous:
git fetch --tagsmake installdropped the.exesuffixfs.read/fs.read_binaryerror detail, plus the directory false-success bug found while testing itcontrib/avcodec+ the contrib-check link wiringEntries are written from the shipped code rather than transcribed from commit subjects, and each keeps the detail that makes it useful in six months — why
ae --versioncannot detect a broken Windows install, why a sandbox refusal needs different wording from an I/O error, why the nightly's stale-toolchain failure reads like a GCC 16 bug and is not.Safety
0deletions. No existing section touched.VERSIONnow matches the newest heading, no stray[current], and none of the four new headings duplicate an existing one.[skip actions]is correct here and bare in the subject: docs-only, zerostd//runtime//compiler/files. It is deliberately not in the PR title, since the title becomes the merge-commit subject and would no-op the release.Worth fixing separately
## [0.435.0]and## [0.497.0]each appear twice onmainalready — untouched here. That is the same pipeline bug from the other direction: the rename can collide with a version heading that already exists. A pipeline that (a) fails loudly when there is no[current]to rename, and (b) refuses to create a duplicate heading, would close both holes. Happy to propose that as its own change.🤖 Generated with Claude Code