Repository navigation
quest: plan a release per shipping merge and pin bumps that follow it - #688
Conversation
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
… host wait Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…on reader Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
Design change: at the round-3 checkpoint, the release quest stops prescribing recovery steps for each way a run can fail. Release now reconciles (Written by Opus 5.5) |
…covery skips Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…ommit Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
Warning Review limit reachedYou've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository. Next included review available in 37 minutes. View limit details
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. |
There was a problem hiding this comment.
✅ No new issues found.
Reviewed changes: the full diff of this quest-only planning PR, plus a search for dangling references to the deleted quests.
pin-bumpsepic: adds the README with its goal, decisions and## Waitingitem for the host wiring.release.md: plans a release per shipping merge, reconcilingmainwith the latest release, tagging last, and keeping one predicate for what ships.discover.mdandfollow-release.md: plan target discovery from the App's installations, and bumps that follow the latest release.- Deleted
m1/release.mdandm1/release-cut.md: nothing links to them any more. The one remaining mention is a pinned-commit permalink. mistake-audit.md: drops the link to the deletedrelease-cut.md.
claude-sonnet-5-5 | 𝕏

Plans the pin-bumps epic: every merge to
mainthat changes what ships becomes a release, and the watcher's bump pass moves every repository that pins quest to it, with no hand bumps.Why: moq-dev/moq sat on
93c9c2efpast #646, so its spawns kept writing XS quests on Haiku 5.5 xhigh and hit rate limits until a hand bump (moq-dev/moq#8623). The bump pass from #550 never ran: the host watcher has no--bumpflags, since that wiring is in the stale draft kixelated/nixos#36. Of the other consumers, moq and web-transport hard-code a revision that nothing moves, moq.pro's Dependabot excludes quest, and the rest move only on Dependabot's weekly run with its 7-day cooldown.[M]Release every merge that changes what ships: Release reconcilesmainwith the latest published release and tags only after the binaries pass their checks. It replacesquest/m1/release.mdandquest/m1/release-cut.md, which this deletes. No release has shipped since v0.1.0.[S]The bump pass finds its own targets: every repository moq-bot is installed on that locks quest.[S]The bump pass follows the latest release.The host side, turning
--bumpon, is planned in kixelated/nixos's tree.Decisions
main).package.json, which conflicts across PRs in the merge queue). This lifts CONTRIBUTING.md's maintainer-only tag rule and the biweekly cut decided on release: bump to 0.2.0 and caveat v0.1.0 #587.mainwith the latest published release on every push and daily, and tags last (rejected: recovery steps for each failure mode, which review kept finding new cases of; semantic-release). Decided at the review's round-3 design checkpoint.--bump, following releases (rejected: Renovate with a stale-stub check, because hosted Renovate can't runquest initorcheck --fixand 48 commits touched skills in 12 days; Renovate with a fixup workflow per consumer; Kixel Code, which runs no host jobs until its m1).--bumplist in the host's config, which went stale within five days).Validation
Quest-only change:
quest checkpasses.Models
Size: S
Implementer: Claude Code, claude-opus-5-5
🤖 Generated with Claude Code
(Written by Opus 5.5)
Review
Rewritten by
quest reviewafter each round.Head: 78a4107
Base: origin/main at a888444
Reviewer: Codex, gpt-6.1-sol, medium effort
Session:
01a1d385-5bed-7f12-bb53-249ead96bb61, round 3Score: 10/10,
The revised plan covers all binary and flake build inputs in a shared release predicate, and the incremental changes introduce no further findings.Rounds: 6; findings 1 (medium 1): fixed 1, refuted 0, open 0
Findings: none
The last round's findings, at a4d4a03:
quest/m0/pin-bumps/release.md:30:Include assets and Nix packaging inputs in the shared predicate;release.md now includes everything the binaries or flake package build from, explicitly names assets/ and nix/, and requires Release's pull request filter to use that same predicate.Reviewer: opencode, opencode-go/deepseek-v4.1-flash, high effort
Session:
ses_e2c7a9f65ffeEy1XAg0I7FCYXt, round 3Score: 10/10,
The head fixes all three remaining findings: the release predicate now names assets/ and nix/, the flake passes the commit and accounts for the older quest-pinned package, and the epic attributes quest-pinned to Dependabot rather than quest check.Rounds: 6; findings 25 (high 2, medium 13, low 10): fixed 24, refuted 1, open 0
Findings: none
The last round's findings, at a4d4a03:
quest/m0/pin-bumps/release.md:31:The new what-ships predicate still omits assets and nix inputs;quest/m0/pin-bumps/release.md:31-35 now defines what ships as everything the binaries or flake package build from, explicitly includes assets/ (bundled at src/cli/skills.ts) and nix/, and says Release's pull-request filter moves to the same predicate.quest/m0/pin-bumps/release.md:51:The version plan never updates the flake.nix call site that instantiates nix/package.nix;quest/m0/pin-bumps/release.md:55-58 now says flake.nix passes the commit to nix/package.nix and notes that quest-pinned builds from the older pinned package.nix, whose call the new argument must not break.quest/m0/pin-bumps/README.md:39:Quest check does not keep the quest-pinned input current;quest/m0/pin-bumps/README.md:39-40 now says quest-pinned stays on Dependabot's weekly nix update (flake.nix), matching .github/dependabot.yml and the flake.nix:11-15 comment.Reviewer: Codex, gpt-6.1-sol, low effort, shadow
Session:
01a1d385-5bed-78a3-a42d-09d37ddc858aRounds: 2; findings 0: fixed 0, refuted 0, open 0