chore(release): bump version to 0.4.4 - #57
Conversation
![claude[bot]](https://avatars.githubusercontent.com/in/1236702?s=60&v=4){kind=small}
There was a problem hiding this comment.
An organization admin can view or raise the cap at claude.ai/admin-settings/claude-code. The cap resets at the start of the next billing period.
Once the cap resets or is raised, reopen this pull request to trigger a review.
There was a problem hiding this comment.
Pull request overview
Cuts the 0.4.4 release by bumping the repository version marker and updating the public landing page + changelog to reflect the new release.
Changes:
- Bump
VERSIONfrom0.4.3to0.4.4. - Update
docs/index.htmldisplayed/versioned release strings and the SHA placeholder to 0.4.4. - Cut
CHANGELOG.mdby moving prior “Unreleased” content under a new0.4.4heading and re-creating an empty “Unreleased” section.
Reviewed changes
Copilot reviewed 3 out of 3 changed files in this pull request and generated no comments.
| File | Description |
|---|---|
| VERSION | Updates the repo’s single-source version marker to 0.4.4. |
| docs/index.html | Updates landing-page visible version strings + release checksum placeholder for 0.4.4. |
| CHANGELOG.md | Creates the 0.4.4 release section and re-establishes an empty Unreleased section. |
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 07cc5a173e
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| <div class="section-inner"> | ||
| <div class="reveal"> | ||
| <div class="section-label">Release 0.4.3</div> | ||
| <div class="section-label">Release 0.4.4</div> |
There was a problem hiding this comment.
Changing this heading to 0.4.4 makes the three cards beneath it appear to describe the new release, but they still summarize 0.4.3: rewrite prompts, search-result copying, and watcher/cold-launch fixes match the 0.4.3 entries in CHANGELOG.md, while the newly cut 0.4.4 section starts with the appearance-picker changes and contains a different set of fixes. Update the cards alongside the release label so the public download page does not attribute the previous release's highlights to 0.4.4.
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
Addressed in 899542d — the three What's new cards are rewritten from the final 0.4.4 changelog section: the Appearance settings tab / status-bar pickers, whole-window light-dark switching (plus the fixed-appearance start-up hang), and the sidebar improvements (halved indent cost, middle elision, rename keeping the extension).
| @@ -1 +1 @@ | |||
| 0.4.3 | |||
| 0.4.4 | |||
There was a problem hiding this comment.
This release commit is titled chore(release): bump version to 0.4.4 and contains no Authored-By: trailer, whereas the repository requires the [<agent>/<workflow>] title form, a non-empty bulleted body, and attribution for the agent that wrote the change. Retitle it and add the required attribution so the release history retains the repository's workflow and authorship provenance.
AGENTS.md reference: AGENTS.md:L25-L28
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
Partially addressed in 490cc7c, partially declined. The Authored-By: claude <agents@vetcoders.io> trailer was present and correct in the original commit — the repo's commit-msg gate (.husky/commit-msg:73-80) derives the agent from the FIRST prefix segment, so claude is the required value and the gate rejects anything else; the commit title also already carried the [claude/worker-basic] form. The one real gap against AGENTS.md was the prose body, which the amend converts to a bulleted list. Tree untouched (empty diff against the pre-amend commit).
|
|
||
| ## [Unreleased] | ||
|
|
||
| ## [0.4.4] - 2026-08-16 |
There was a problem hiding this comment.
Inserting the 0.4.4 heading here places every formerly Unreleased entry through line 491 into this release, including the filesystem-watcher rescan fix at line 489. The same single-flight/cancellation fix is already explicitly recorded under 0.4.3 at line 508, so release-note consumers now see one change attributed to both versions. Remove the stale duplicate from the new section, or otherwise preserve it under the release in which it shipped.
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
Addressed in ddaec47 — the watcher rescan bullet is removed from 0.4.4 (same root cause and fix as 0.4.3's "Workspace watcher scan pile-up"). A full sweep compared every 0.4.4 bullet that predates the 0.4.3 cut (106 of 120) against the 0.4.3 section: this was the only true duplicate; five partial overlaps were examined and kept because each carries a distinct fix (details in the commit body).
- Cut the Unreleased changelog section into `## [0.4.4] - 2026-08-16`, mirroring the mechanical 0.4.2-style cut (90fe5b2): Unreleased stays as an empty header and its full prior content moves down under the new version. - Carry the stray "Release provenance verification..." entry along in that move: it never rode with 0.4.3's hand-authored notes, so 0.4.4 is its first real release home. - Bump `VERSION` to 0.4.4. - Update the `docs/index.html` version and SHA-placeholder slots to match. Authored-By: claude <agents@vetcoders.io> session_id: 01a0057d-7dd8-7421-8363-844e9cf10946 time: 2026-08-16T00:47:29+02:00 runtime: iterm2
…rom 0.4.4 - Remove the "Filesystem-watcher rescan storm on busy workspace roots" bullet from `## [0.4.4]`: it describes the same single-flight/cancellation fix that already shipped as "Workspace watcher scan pile-up" under `## [0.4.3]`, so 0.4.4 was re-announcing a released fix. - Swept every other 0.4.4 bullet that predates the 0.4.3 cut (`git show 46372fd^:CHANGELOG.md`) against the 0.4.3 section; this was the only semantic duplicate. The remaining overlaps (isolated UI smoke, toolbar overflow, temporary Application Support roots) each carry substance 0.4.3 never described, so they stay. - "Release provenance verification..." is untouched: 0.4.3 does not describe it. Authored-By: claude <agents@vetcoders.io> session_id: 01a0057d-7dd8-7421-8363-844e9cf10946 time: 2026-08-16T01:40:41+02:00 runtime: iterm2
…or 0.4.4 - Replace the three "What's new" cards, which still described 0.4.3's highlights (per-intent rewrite prompts, Copy for Agent, the watcher/cold- launch work) under a 0.4.4 heading. - Write three cards from the final 0.4.4 changelog section instead: the theme and flavor pickers moving to the status-bar chip plus the new Settings > Appearance tab; the light/dark flip landing on the whole window at once and no longer hanging start-up on a large document; and the sidebar's halved indent, middle elision and extension-preserving rename. - Cards only — version slots, checksum slot and every other section untouched. - Gate: `./scripts/test-landing-page.sh` 66 passed, 0 failed. Authored-By: claude <agents@vetcoders.io> session_id: 01a0057d-7dd8-7421-8363-844e9cf10946 time: 2026-08-16T01:42:15+02:00 runtime: iterm2
07cc5a1 to
899542d
Compare
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 3 out of 3 changed files in this pull request and generated no new comments.
Suppressed comments (1)
CHANGELOG.md:489
- The changelog cut dropped the “Filesystem-watcher rescan storm on busy workspace roots” entry entirely. The PR description indicates everything under
[Unreleased]should have been moved into the0.4.4section, so this looks like an accidental deletion; please re-add the bullet so the 0.4.4 notes are complete.
- Released app binaries no longer carry absolute `LC_RPATH` entries. The linker baked the builder's `Vendor/qube-ffi/<profile>` checkout path (plus the Xcode toolchain path) into every shipped binary, and dyld searched those _before_ `@executable_path/../Frameworks` — so on a machine where the builder's path happened to exist, the app loaded a qube-ffi dylib from outside its own bundle. The release pipeline now strips every absolute rpath before signing, refuses to sign or package a bundle where one survives, and verifies that each remaining `@rpath` dependency still resolves inside the bundle.
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 899542df0d
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
|
|
||
| ## [Unreleased] | ||
|
|
||
| ## [0.4.4] - 2026-08-16 |
There was a problem hiding this comment.
Fresh evidence beyond the removed watcher duplicate is that the 0.4.3 release commit 0c2a1d5 already had VERSION set to 0.4.3 while retaining the entries that now begin under this release's second ### Added heading, including Recovered Drafts and deterministic New behavior. Moving the entire stale Unreleased block under this heading therefore attributes hundreds of changes shipped in 0.4.3 to 0.4.4; preserve only changes made after the 0.4.3 release and assign the older block to its actual release.
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
Declined on evidence: 0.4.3 was never released. gh release list shows the latest published release is v0.4.2 (2026-07-22) — there is no v0.4.3 tag, no GitHub release, and the landing page's checksum slot still holds the never-stamped placeholder. 0c2a1d5 is an ordinary PR merge (#48), not a release commit; VERSION reading 0.4.3 in-tree is a working label, not shipping. For a user on the latest actual release (0.4.2), everything now under 0.4.4 ships for the first time in 0.4.4, so the attribution is correct. The 0.4.3 section stays as recorded history of that internal cut; the one genuine double-listing (the watcher fix) was already removed in ddaec47.
| <p>Workspace scans no longer stack up under heavy filesystem churn — one walk at a time, cancelled cleanly when outdated. Window handling owns the cold launch, and an empty workspace keeps the launcher alive instead of quitting.</p> | ||
| <span class="tagchip">Sidebar</span> | ||
| <h3>Deep files, whole names</h3> | ||
| <p>A nesting level costs half the indent it used to, and every row, search hit and path now elides from the middle — so a deeply nested, date-prefixed file keeps both its date and the suffix that tells it apart. Renaming a file no longer drops its extension.</p> |
There was a problem hiding this comment.
Fresh evidence in the revised card is the new extension-renaming claim: that fix is already present in the changelog tree of the 0c2a1d5 release commit, where VERSION is 0.4.3. Consequently, visitors reading the 0.4.4 “What’s new” card are still told that an already-shipped 0.4.3 fix is new in this release; replace this sentence with a sidebar change introduced after that release.
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
Declined on the same evidence as the CHANGELOG thread: no 0.4.3 was ever published (latest actual release is v0.4.2 — no v0.4.3 tag or GitHub release, landing page never stamped), so the rename-extension fix has never reached a user and is genuinely new in 0.4.4. The card stands.
Cut the release version to 0.4.4 (operator decision).
Gate: `./scripts/test-landing-page.sh` 66/66, run before and after the CHANGELOG cut.
Supersedes #56 (auto-closed when its stacked base `fix/pensieve-build-keychain-unlock` was deleted on the #55 merge).
🤖 Generated with Claude Code