Conversation
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
![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: 435b7c55cc
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".
| [package] | ||
| name = "fledge" | ||
| version = "1.7.2" | ||
| version = "1.8.0" |
There was a problem hiding this comment.
When the v1.8.0 tag is pushed, .github/workflows/release.yml only tests, builds binaries, and creates the GitHub release; it never checks out with a Rust toolchain in the release job or runs cargo publish. The post-release workflow merely warns when crates.io lags, despite CONTRIBUTING.md stating that release.yml publishes there. Consequently, the documented cargo install fledge channel remains on 1.7.2 after this release unless a maintainer performs an undocumented manual publication.
Useful? React with 👍 / 👎.
| The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/), | ||
| and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html). | ||
|
|
||
| ## [v1.8.0] - 2026-09-18 |
There was a problem hiding this comment.
Cutting this release inserts the v1.8.0 section ahead of the existing Unreleased section without consuming it. The two lifecycle-hook entries at lines 60–61 describe commit c4abcb2, which is already included in v1.8.0 at line 41, so the published changelog now labels shipped functionality as unreleased and will continue carrying those stale entries into later releases.
Useful? React with 👍 / 👎.
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
There was a problem hiding this comment.
❌ Corvin says...
_
<(;\ .oO(oh no...)
|/(\
\(\\
" "\\
"Caw... validation failed..."
CI Summary
| Check | Status |
|---|---|
| Dependency Audit | ✅ Passed |
| Integration (3 OS) | ❌ skipped |
| Lint (fmt + clippy) | ✅ Passed |
| Spec Validation | ✅ Passed |
| Tests (3 OS) | ❌ failure |
Powered by corvid-pet
435b7c5 to
02e7767
Compare
Superseded by updated review.
There was a problem hiding this comment.
❌ Corvin says...
_
<(;\ .oO(oh no...)
|/(\
\(\\
" "\\
"Even the dumpster of code seems empty today."
CI Summary
| Check | Status |
|---|---|
| Dependency Audit | ✅ Passed |
| Integration (3 OS) | ❌ skipped |
| Lint (fmt + clippy) | ✅ Passed |
| Spec Validation | ✅ Passed |
| Tests (3 OS) | ❌ failure |
Powered by corvid-pet
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 02e77675fa
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".
| [package] | ||
| name = "fledge" | ||
| version = "1.7.2" | ||
| version = "1.8.0" |
There was a problem hiding this comment.
Fresh evidence relative to the earlier review is that parent commit 7214df6 now adds an unconditional cargo publish step to .github/workflows/release.yml, while this release commit states that 1.8.0 was already manually published and verified on crates.io. When the v1.8.0 tag fires, crates.io will reject that upload—the official Cargo documentation states that a published version can never be overwritten—so the publish job exits non-zero before its registry-verification step and every rerun remains red. Either avoid prepublishing this version or make the job verify and accept the exact already-published version before attempting the upload.
AGENTS.md reference: AGENTS.md:L361-L363
Useful? React with 👍 / 👎.
| The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/), | ||
| and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html). | ||
|
|
||
| ## [v1.8.0] - 2026-09-18 |
There was a problem hiding this comment.
The reviewed parent is 7214df6 (Fix: publish to crates.io from the release workflow, and verify the registry (#531)), but the new v1.8.0 section enumerates the other 22 post-v1.7.2 commit hashes and omits this direct parent. Consequently the release ships a significant change to its distribution process without recording it in the changelog; regenerate the section from the actual parent or add this fix explicitly.
Useful? React with 👍 / 👎.
corvid-agent
left a comment
corvid-agent
left a comment
There was a problem hiding this comment.
CoS REQUEST_CHANGES — 02e77675fa4b
Release notes / Cargo bump for v1.8.0 look coherent. Linux + macOS test, trust, lint, audit, intent/spec-check, and the Action install matrix are green. windows is refused with a readable message also passes.
Only red: test (windows-latest) — default_temp_env_points_github_at_a_dead_port expects a specific "refused" string and gets Windows os error 10061 wording instead. Per standing Linux-first (no Windows target unless Leif asks), prefer either:
- broaden the isolation assert to accept the Windows refused wording, or
- drop/skip that Windows matrix cell for the release.
Happy to APPROVE once that sole red is cleared (or explicitly waived by Leif). No merge from CoS.
…#535) default_temp_env_points_github_at_a_dead_port recognized a refused connection only by the words "connection refused". ureq 3.3.0 synthesized exactly that message on every platform. ureq 3.4 passes the OS error through instead, so Windows now reports WSAECONNREFUSED's "No connection could be made because the target machine actively refused it. (os error 10061)" and the test failed on the v1.8.0 release PR (#532), whose lockfile carries ureq 3.4.2. Accept "actively refused" and "os error 10061" as well. The loopback assertion on the child's GitHub base is unchanged, and a generic "failed" still does not count, so a real api.github.com failure still fails the test. Adds the SpecSync change record (draft, awaiting definition approval). Claude-Session: https://claude.ai/code/session_01V3ZZAEiUP7xRJPozhZb6rL Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
|
Update (17:5xZ): #535 merged (260063a). This PR was closed by the 0xLeif account at 17:19Z. Not knowing about the close, the agent then force-pushed the rebase described below to this branch: Status from orc's CI-fix agent. Nothing on this branch has changed yet. corvid-agent's CHANGES_REQUESTED (the only red, Once #535 lands, this PR gets rebased (it applies cleanly) with these changelog fixes, which also cover both Codex P2s:
Needs a maintainer decision before tagging (Codex P1): crates.io already has fledge 1.8.0 (published 2026-09-19).
Options: (a) ship this as v1.8.0 and accept both; (b) tag 🤖 Generated with Claude Code |
|
Superseded by #536 (v1.8.1), per Leif's decision: 1.8.0 is already on crates.io from 🤖 Generated with Claude Code |
2 participants
Summary
Bumps to 1.8.0 —
Cargo.toml,Cargo.lock,flake.nix— with a changelog generated across all 22 commits since v1.7.2.max_version: 1.8.0, not yanked. So the registry currently leads the repo, and this PR plus the tag bring them back in line.Why the registry went first
crates.io was stuck at 1.7.0 while v1.7.1 and v1.7.2 were both tagged — two releases cut and never shipped, because
release.ymlhad no publish step and nothing said so (see #531). Publishing 1.8.0 closes that gap by supersession; 1.7.1 and 1.7.2 stay GitHub-only, which is harmless since crates.io does not require contiguous versions.What's in 1.8.0
Features:
--streamfor live task output (#509),fledge spec lint(#505), the root-levelaction.ymlGitHub Action (#511).Fixes: the diamond-DAG dependency bug (#513), the task-graph depth bound that turned a stack-overflow abort into a real error (#525), Astro 5→7 clearing ten Dependabot advisories (#528), two lifecycle-hook fixes (#520, #521).
Plus the test-isolation work (#504) and the full SDD ledger closeout (#526, #527, #529, #530).
Verification
fledge lanes run pre-commit— green (3 steps, 1m25s)fledge spec check— 33 specs, 0 errors, 0 warningscargo publish --dry-run— packaged 709 files, verified and compiled as 1.8.0cargo publish— succeededmax_version: 1.8.0,yanked: false. Not inferred from cargo's exit codeAfter merge
Tag
v1.8.0on main.release.ymlbuilds the artifacts and creates the GitHub release; its publish job (#531, if merged first) will find 1.8.0 already present.The stale
chore/release-1-8-0branch was deleted — it had been cut atf659d69, so its changelog predated #524 through #530.🤖 Generated with Claude Code