chore: add a formatter - #64
Merged
Merged
Conversation
singleQuote=false was chosen by counting this repo's own imports, not by fleet decree. The fleet is genuinely split and the two repos that already had a .prettierrc disagreed with each other, so there was no standard to restore. Quote style does not cross repo boundaries; having a gate does. Markdown is ignored for now — prettier rewraps prose, which would bury the real diff. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Mechanical. No behaviour change. This SHA is listed in .git-blame-ignore-revs so `git blame` skips it. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
catomean
force-pushed
the
chore/add-a-formatter
branch
from
August 29, 2026 09:39
7cb158a to
be645f1
Compare
The previous commit regenerated pnpm-lock.yaml with local pnpm 11, which re-resolved the `github:catomean/ai-forms` shorthand from a codeload tarball to git+ssh. CI runs pnpm 9, which cannot read that entry — and could not fetch over SSH even if it could, having no key. The lockfile was valid on the machine that wrote it and broken everywhere else. Resolved with pnpm 9 instead. The rollout script now derives the major from packageManager/PNPM_VERSION and refuses any lockfile that gains an ssh-form resolution.
pnpm records whichever transport it could use when the lockfile was written. This laptop has an SSH key, so pnpm wrote git+ssh; CI has none and the install died on 'Permission denied (publickey)'. Resolving with GIT_SSH_COMMAND=/bin/false reproduces CI's capabilities and yields git+https, which needs no credentials for a public repo.
kivvi already depended on prettier, so this PR is script wiring, not a new dependency — and it should not have rewritten the lockfile at all. Regenerating it re-resolved the `github:catomean/ai-forms` shorthand three different ways, none of which CI could use: git+ssh (no key on the runner), then git+https (fetches, but pnpm's virtual-store path for a git dep failed to import). Only main's original codeload tarball works. The resolution recorded in a lockfile depends on what the machine writing it could reach, which makes 'it resolved fine locally' worth nothing. So: restore main's lockfile and change only the specifier, which is what --frozen-lockfile actually compares, and format with 3.8.0 — the version this repo's own lockfile already pinned — rather than dragging it to the fleet's 3.9.6 and reformatting against a version it does not install.
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.
3 of 30 repos had any formatter at all. This adds one here, and wires
format:checkintoverifyso it actually gates.Why this repo's own style, and not a fleet standard
There was never a standard to restore. Measured across the fleet: excluding openclaw, ~28k single-quoted imports against ~15k double-quoted — and the two repos that already had a
.prettierrcdisagreed with each other (trailingComma: allvses5,arrowParens: alwaysvsavoid).Forcing one quote style would rewrite every string in roughly half the fleet for no benefit. And by the test this whole exercise runs on — does the choice cross repo boundaries? — quote style does not. Nobody copies a file between repos expecting the quotes to match.
So
singleQuotehere was chosen by counting this repo's own imports. The uniformity that matters is that a gate exists everywhere, not that the gate says the same thing everywhere.What is deliberately out of scope
.prettierignoreturns it on later.Reading the diff
Three commits, on purpose:
.git-blame-ignore-revslisting commit 2Run
git config blame.ignoreRevsFile .git-blame-ignore-revsonce andgit blamewill skip straight past the reformat.🤖 Generated with Claude Code
https://claude.ai/code/session_01UvjGNAS9CMfEGNW26tUR4P