replace micromatch with picomatch

[beeequeue]

currently stuck on the default config no longer parsing

[changeset-bot]

🦋 Changeset detected

Latest commit: a8e32c2

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 7 packages

Name	Type
@changesets/config	Patch
@changesets/git	Patch
@changesets/apply-release-plan	Patch
@changesets/cli	Patch
@changesets/get-release-plan	Patch
@changesets/read	Patch
@changesets/release-utils	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[socket-security]

Review the following changes in direct dependencies. Learn more about Socket for GitHub.

Diff	Package	Supply Chain Security	Vulnerability	Quality	Maintenance	License
	@​types/​picomatch@​4.0.3
	dotenv@​16.6.1

View full report

[Andarist]

I wonder, how does it compare to the builtin? https://nodejs.org/docs/latest-v22.x/api/fs.html#fsglobpattern-options-callback

[beeequeue]

picomatch wayyyy faster iirc as the builtins use minimatch

but i don't have any benchmarks saved so i'll try to find or make some

i also mentioned this on bsky recently :^)

[bluwy]

We need to use this instead so we don't touch the fs: https://nodejs.org/api/path.html#pathmatchesglobpath-pattern

Open to trying it though but we still kinda need a wrapper to handle array of files and patterns.

[Andarist]

It would be nice to know the benchmark results. However, that wouldn't really be a bottleneck in the case of Changesets - so we could just favor less dependencies over a more performans library here.

[codecov]

Codecov Report

❌ Patch coverage is 91.66667% with 3 lines in your changes missing coverage. Please review.
✅ Project coverage is 81.99%. Comparing base (ed6728c) to head (a8e32c2).
⚠️ Report is 1 commits behind head on next.

Files with missing lines	Patch %	Lines
packages/git/src/index.ts	78.57%	2 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             next    #1953      +/-   ##
==========================================
+ Coverage   81.90%   81.99%   +0.09%     
==========================================
  Files          54       54              
  Lines        2432     2461      +29     
  Branches      706      718      +12     
==========================================
+ Hits         1992     2018      +26     
- Misses        398      400       +2     
- Partials       42       43       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[bluwy]

I updated getUnmatchedPatterns that wasn't migrated quite right, and I updated the matching to handle this specific behavior from micromatch:

import picomatch from 'picomatch'
import micromatch from 'micromatch'

const pkgs = ['pkg-a', 'pkg-b', '@scope/pkg-c', '@scope/pkg-d']
const globs = ['pkg-*', '!pkg-b', '@scope/*']

const matcher = picomatch(globs)
console.log('picomatch:', pkgs.filter((pkg) => matcher(pkg)))
// => [ 'pkg-a', 'pkg-b', '@scope/pkg-c', '@scope/pkg-d' ]

console.log('micromatch:', micromatch(pkgs, globs))
// => [ 'pkg-a', '@scope/pkg-c', '@scope/pkg-d' ]

The negation is order-sensitive. picomatch simply returns true if any pattern returns true.

[Andarist]

Hm, shouldn't we inherit picomatch semantics though? We do mention what matcher library gets used under the hood - so people can reason about their glob patterns etc. Diverging from that could feel surprising.

[beeequeue]

i would prefer leaving the behavior to picomatch, less code for us to worry about

[bluwy]

If we leave it entirely to picomatch, they'd have to craft every pattern to not include it, like "pkg-!(a)" instead to get the negation pattern work like before. I feel like this will be quite difficult to migrate though and not worth it. For reference, here's how picomatch implements array support.

Should we consider documenting instead? I also think the link to picomatch also mostly refers to its pattern syntax than the array matching behaviour.

[Andarist]

If we'd decide to use the builtin node API then this discussion would become obsolete :p I'm fine either way, so I'll leave you both to decide if we should use the node API or picomatch.

[bluwy]

If path.matchesGlob works and is compatible, I think we should use it. I'm just a bit lazy to check and update it 😄