Skip to content

docs(changelog): fold Unreleased into 0.7.0, and write up the thirty PRs it was missing - #358

Merged
rainmanjam merged 1 commit into
mainfrom
docs/changelog-0.7.0
Aug 15, 2026
Merged

docs(changelog): fold Unreleased into 0.7.0, and write up the thirty PRs it was missing#358
rainmanjam merged 1 commit into
mainfrom
docs/changelog-0.7.0

Conversation

@rainmanjam

Copy link
Copy Markdown
Owner

Found while checking for deltas before tagging v0.7.0. Two problems, and the first would have shipped a release whose own changelog contradicted its security advisories.

1. [Unreleased] sat above [0.7.0]

Keep a Changelog puts newest first, so its contents were newer than the release below them. What was sitting there: the stream-key sealing, the scrub fixes, and Twitch Enhanced Broadcasting — precisely what GHSA-7jqx-76vq-hvfc names as "fixed in 0.7.0".

Tagging as it stood would have published an advisory pointing at a release whose changelog did not mention the fix.

2. The file stopped at issue #326

Verified by concept rather than by number, because entries cite issues and not pull requests — a PR-number check reports a false 60/60 missing:

Concept Occurrences before
maxrate / bufsize (#341) 0
fps_mode (#342) 0
"strength meter" (#346) 0
secret.key (#347) 0
"Rumble" (#312) 0

Roughly thirty merged PRs, including a data-loss guard on upgrade an operator needs to read before upgrading.

Checking below #327 for the same reason found two more: the services platform registry, and the five live acceptance suites — a headline of this release, since each of the first three found a defect on its first live run that no unit test could reach.

What this does

Net: 70 bullets in 0.7.0, up from 45.

The prose is moved, not retyped

The fold is done programmatically and asserts:

old lines dropped:                 0
0.6.0-and-older byte-identical:    True
malformed links:                   0

Both assertions hold. Nothing in the existing changelog was rewritten by hand, so nothing could be silently reworded.

One correction worth recording

Five entries cited /issues/N for what are actually pull requests. My first check used gh issue view, which resolves a pull request happily — so it passed. The check that caught it asks the API whether the object carries a pull_request field. All 33 links in the diff are now verified against that.

Still open, and stated in the file rather than omitted

Nothing proves an OAuth refresh succeeds, and no chat suite performs a valid login. Both steps are written and skipping until an account is connected, so the hour-four token failure is bounded from one side only. The Testing section says so.

https://claude.ai/code/session_01HeLrWaDmsNeeNSbHQfEofX

…PRs it was missing

Two problems, and the first would have shipped a release whose own changelog
contradicted its security advisories.

`[Unreleased]` sat ABOVE `## [0.7.0] - 2026-08-14`, which under Keep a
Changelog means its contents are NEWER than the release below it. What was
sitting there was the stream-key sealing, the scrub fixes and Twitch Enhanced
Broadcasting -- exactly the work GHSA-7jqx-76vq-hvfc names as "fixed in
0.7.0". Tagging as it stood would have published an advisory pointing at a
release whose changelog did not mention the fix.

And the file stopped at issue #326. Everything after was undocumented --
verified by concept rather than by number, because entries cite issues and not
pull requests: `maxrate`, `bufsize`, `fps_mode`, "strength meter",
`secret.key` and "Rumble" each appeared zero times. That is roughly thirty
merged pull requests, including a data-loss guard on upgrade that an operator
needs to read BEFORE upgrading, not after.

Checking the range below #327 for the same reason found two more gaps: the
services platform registry, and the five live acceptance suites -- which are a
headline of this release, since each of the first three found a defect on its
first live run that no unit test could reach.

The existing prose is MOVED, not retyped: the fold is done programmatically
and asserts that no line of the old file is absent from the new one, and that
0.6.0 and everything older is byte-identical. Both assertions hold.

Also corrected five links that named /issues/N for what are actually pull
requests. `gh issue view` resolves a pull request happily, so the first check
passed and the second one -- asking the API whether the object has a
`pull_request` field -- is the one that caught it.

Claude-Session: https://claude.ai/code/session_01HeLrWaDmsNeeNSbHQfEofX
Copilot AI lite review requested due to automatic review settings August 14, 2026 22:03
@sonarqubecloud

Copy link
Copy Markdown

Copilot AI left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull request overview

Updates the project changelog in preparation for tagging v0.7.0, ensuring the release notes accurately reflect fixes (including security-related items) and the full set of merged work included in the release.

Changes:

  • Moves the prior [Unreleased] Security/Added entries into the [0.7.0] section and leaves [Unreleased] as “Nothing yet.”
  • Expands the [0.7.0] notes to cover the previously missing set of items (#327#357) across multiple categories (Security/Added/Changed/Fixed/Testing).
  • Updates/aligns the version comparison links so [Unreleased] now compares from v0.7.0.

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

@rainmanjam
rainmanjam merged commit 9a50141 into main Aug 15, 2026
29 checks passed
@rainmanjam
rainmanjam deleted the docs/changelog-0.7.0 branch August 16, 2026 19:51
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.

2 participants