fs: cache glob matcher functions in path.matchesGlob

[beeequeue]

this PR adds benchmarks for path.matchesGlob() as well as a simple cache for pattern matching functions.

rationale

most glob matching libraries in the ecosystem (including minimatch) provide an API to let users re-use matchers instead of re-creating them every call.

node doesn't provide such an API, and does not re-use matchers when possible, leading to it being much slower (10-15x) than just using new Minimatch() directly when matching a pattern more than once.

implementation details

i decided to "hide" away the cache inside the function rather than add a new API as most users will expect the function to not be a potential footgun, and the pattern has been used before in zeptomatch

since matchGlobPattern() doesn't accept any options other than pattern and path we can safely cache it by the pattern string. if this changes in the future the cache would need to accomodate the new options in the cache keys.

i went with the simplest Map cache implementation i could come up with as i couldn't find any LRU implementations anywhere in the codebase, if you have any suggestions for improvements please let me know!

testing

i'm not sure how to test this properly, e.g. if i should export the cache instance and check its size, etc

some guidance here is very appreciated :)

---

benchmark results

ran on my windows 11 machine with an AMD Ryzen 9800X3D

note: these benchmarks do not test the case where we evict keys from the cache

 confidence improvement accuracy (*) (**) (***)
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1013.11 % ±23.55% ±31.74% ±42.14%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.test.ts' *** 2005.02 % ±48.08% ±64.80% ±86.02%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 2073.47 % ±27.53% ±37.10% ±49.24%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 869.89 % ±18.72% ±25.22% ±33.48%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.test.ts' *** 1726.85 % ±29.66% ±39.97% ±53.07%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 1777.43 % ±25.94% ±34.95% ±46.40%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1074.40 % ±22.04% ±29.70% ±39.43%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.test.ts' *** 1424.64 % ±28.72% ±38.70% ±51.38%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 1460.18 % ±17.52% ±23.60% ±31.33%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 659.77 % ±7.00% ±9.43% ±12.52%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 1056.32 % ±12.77% ±17.21% ±22.84%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 492.15 % ±4.26% ±5.73% ±7.60%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\index.ts' *** 858.27 % ±6.41% ±8.64% ±11.46%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 565.03 % ±5.60% ±7.54% ±10.01%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 915.51 % ±9.18% ±12.38% ±16.42%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 431.51 % ±3.11% ±4.18% ±5.51%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\index.ts' *** 746.71 % ±6.50% ±8.75% ±11.62%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 597.13 % ±12.85% ±17.31% ±22.97%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 745.44 % ±12.25% ±16.50% ±21.90%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 421.16 % ±3.44% ±4.61% ±6.08%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\index.ts' *** 594.88 % ±10.85% ±14.62% ±19.41%

i also have a personal benchmarking repo where i found this issue, where the results now look like this:

screenshot

[nodejs-github-bot]

Review requested:

• @nodejs/performance

[beeequeue]

i re-used createMatcher from above as i saw that it uses the same default options as the previous lines did.

i thought about putting the caching in that function instead, but then we would need to base the caching on all the options it accepts, so i decided to keep it here.

[beeequeue]

@avivkeller@MoLow if you have time, could you take a look at this PR? 🥰

[beeequeue]

@avivkeller is there anything i can do to help get this merged?

[avivkeller]

cc @nodejs/fs PTAL

[mcollina]

lgtm

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74794/

[beeequeue]

@mcollina should i do something about these failing checks? i don't have access to see the results and i'm unsure if any of the test failures i see on my machine are because of my changes

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74831/

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74839/

[nodejs-github-bot]

Commit Queue failed

- Loading data for nodejs/node/pull/63915
✔ Done loading data for nodejs/node/pull/63915
----------------------------------- PR info ------------------------------------
Title fs: cache glob matcher functions in path.matchesGlob (#63915)
Author Adam Haglund <adam@haglund.dev> (@beeequeue, first-time contributor)
Branch beeequeue:matchesGlob-perf -> nodejs:main
Labels fs, author ready
Commits 2
- path: add benchmarks for path.matchesGlob()
- fs: add pattern cache for matchGlobPattern()
Committers 1
- bq <adam@haglund.dev>
PR-URL: https://github.com/nodejs/node/pull/63915
Reviewed-By: Aviv Keller <me@aviv.sh>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
------------------------------ Generated metadata ------------------------------
PR-URL: https://github.com/nodejs/node/pull/63915
Reviewed-By: Aviv Keller <me@aviv.sh>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
--------------------------------------------------------------------------------
ℹ This PR was created on Sun, 14 Jun 2026 13:41:33 GMT
✔ Approvals: 2
✔ - Aviv Keller (@avivkeller): https://github.com/nodejs/node/pull/63915#pullrequestreview-4630220976
✔ - Matteo Collina (@mcollina) (TSC): https://github.com/nodejs/node/pull/63915#pullrequestreview-4688113820
✔ Last GitHub CI successful
ℹ Last Full PR CI on 2026-07-15T15:29:50Z: https://ci.nodejs.org/job/node-test-pull-request/74839/
- Querying data for job/node-test-pull-request/74839/
✔ Build data downloaded
✔ Last Jenkins CI successful
--------------------------------------------------------------------------------
✔ No git cherry-pick in progress
✔ No git am in progress
✔ No git rebase in progress
--------------------------------------------------------------------------------
- Bringing origin/main up to date...
From https://github.com/nodejs/node
* branch main -> FETCH_HEAD
✔ origin/main is now up-to-date
- Downloading patch for 63915
From https://github.com/nodejs/node
* branch refs/pull/63915/merge -> FETCH_HEAD
✔ Fetched commits as 874e96ee4cd0..f87f628b4e4b
--------------------------------------------------------------------------------
[main d9c89a650e] path: add benchmarks for path.matchesGlob()
Author: bq <adam@haglund.dev>
Date: Sun Jun 14 14:53:03 2026 +0200
2 files changed, 57 insertions(+)
create mode 100644 benchmark/path/matchesGlob-posix.js
create mode 100644 benchmark/path/matchesGlob-win32.js
[main a3e8124ad7] fs: add pattern cache for matchGlobPattern()
Author: bq <adam@haglund.dev>
Date: Sat Jun 13 18:06:10 2026 +0200
1 file changed, 20 insertions(+), 10 deletions(-)
✔ Patches applied
There are 2 commits in the PR. Attempting autorebase.
(node:395) [DEP0190] DeprecationWarning: Passing args to a child process with shell option true can lead to security vulnerabilities, as the arguments are not escaped, only concatenated.
(Use `node --trace-deprecation ...` to show where the warning was created)
Rebasing (2/4)
Executing: git node land --amend --yes
--------------------------------- New Message ----------------------------------
path: add benchmarks for path.matchesGlob()
Signed-off-by: Adam Haglund <adam@haglund.dev>

PR-URL: #63915

Reviewed-By: Aviv Keller <me@aviv.sh>

Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
[detached HEAD 39c08b796c] path: add benchmarks for path.matchesGlob()

Author: bq <adam@haglund.dev>

Date: Sun Jun 14 14:53:03 2026 +0200

2 files changed, 57 insertions(+)

create mode 100644 benchmark/path/matchesGlob-posix.js

create mode 100644 benchmark/path/matchesGlob-win32.js

Rebasing (3/4)

Rebasing (4/4)

Executing: git node land --amend --yes

--------------------------------- New Message ----------------------------------

fs: add pattern cache for matchGlobPattern()
Signed-off-by: Adam Haglund <adam@haglund.dev>

PR-URL: #63915

Reviewed-By: Aviv Keller <me@aviv.sh>

Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
[detached HEAD a1983d94de] fs: add pattern cache for matchGlobPattern()

Author: bq <adam@haglund.dev>

Date: Sat Jun 13 18:06:10 2026 +0200

1 file changed, 20 insertions(+), 10 deletions(-)

Successfully rebased and updated refs/heads/main.
ℹ Add commit-queue-squash label to land the PR as one commit, or commit-queue-rebase to land as separate commits.

https://github.com/nodejs/node/actions/runs/29491987867

[nodejs-github-bot]

Landed in 1d4e86e...b8f81e9

[Edgarruiz8585]

this PR adds benchmarks for path.matchesGlob() as well as a simple cache for pattern matching functions.

rationale

most glob matching libraries in the ecosystem (including minimatch) provide an API to let users re-use matchers instead of re-creating them every call.

node doesn't provide such an API, and does not re-use matchers when possible, leading to it being much slower (10-15x) than just using new Minimatch() directly when matching a pattern more than once.

implementation details

i decided to "hide" away the cache inside the function rather than add a new API as most users will expect the function to not be a potential footgun, and the pattern has been used before in zeptomatch

since matchGlobPattern() doesn't accept any options other than pattern and path we can safely cache it by the pattern string. if this changes in the future the cache would need to accomodate the new options in the cache keys.

i went with the simplest Map cache implementation i could come up with as i couldn't find any LRU implementations anywhere in the codebase, if you have any suggestions for improvements please let me know!

testing

i'm not sure how to test this properly, e.g. if i should export the cache instance and check its size, etc

some guidance here is very appreciated :)

---

benchmark results

ran on my windows 11 machine with an AMD Ryzen 9800X3D

note: these benchmarks do not test the case where we evict keys from the cache

 confidence improvement accuracy (*) (**) (***)
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1013.11 % ±23.55% ±31.74% ±42.14%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.test.ts' *** 2005.02 % ±48.08% ±64.80% ±86.02%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 2073.47 % ±27.53% ±37.10% ±49.24%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 869.89 % ±18.72% ±25.22% ±33.48%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.test.ts' *** 1726.85 % ±29.66% ±39.97% ±53.07%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 1777.43 % ±25.94% ±34.95% ±46.40%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1074.40 % ±22.04% ±29.70% ±39.43%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.test.ts' *** 1424.64 % ±28.72% ±38.70% ±51.38%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 1460.18 % ±17.52% ±23.60% ±31.33%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 659.77 % ±7.00% ±9.43% ±12.52%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 1056.32 % ±12.77% ±17.21% ±22.84%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 492.15 % ±4.26% ±5.73% ±7.60%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\index.ts' *** 858.27 % ±6.41% ±8.64% ±11.46%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 565.03 % ±5.60% ±7.54% ±10.01%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 915.51 % ±9.18% ±12.38% ±16.42%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 431.51 % ±3.11% ±4.18% ±5.51%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\index.ts' *** 746.71 % ±6.50% ±8.75% ±11.62%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 597.13 % ±12.85% ±17.31% ±22.97%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 745.44 % ±12.25% ±16.50% ±21.90%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 421.16 % ±3.44% ±4.61% ±6.08%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\index.ts' *** 594.88 % ±10.85% ±14.62% ±19.41%

also have a personal benchmarking repo where i found this issue, where the results now look like this:
Edgarruiz8585

screenshot

[Edgarruiz8585]

Verificado por Edgarruiz8585

[beeequeue]

this PR adds benchmarks for path.matchesGlob() as well as a simple cache for pattern matching functions.

rationale

most glob matching libraries in the ecosystem (including minimatch) provide an API to let users re-use matchers instead of re-creating them every call.

node doesn't provide such an API, and does not re-use matchers when possible, leading to it being much slower (10-15x) than just using new Minimatch() directly when matching a pattern more than once.

implementation details

i decided to "hide" away the cache inside the function rather than add a new API as most users will expect the function to not be a potential footgun, and the pattern has been used before in zeptomatch

since matchGlobPattern() doesn't accept any options other than pattern and path we can safely cache it by the pattern string. if this changes in the future the cache would need to accomodate the new options in the cache keys.

i went with the simplest Map cache implementation i could come up with as i couldn't find any LRU implementations anywhere in the codebase, if you have any suggestions for improvements please let me know!

testing

i'm not sure how to test this properly, e.g. if i should export the cache instance and check its size, etc

some guidance here is very appreciated :)

---

benchmark results

ran on my windows 11 machine with an AMD Ryzen 9800X3D

note: these benchmarks do not test the case where we evict keys from the cache

 confidence improvement accuracy (*) (**) (***)
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1013.11 % ±23.55% ±31.74% ±42.14%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.test.ts' *** 2005.02 % ±48.08% ±64.80% ±86.02%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 2073.47 % ±27.53% ±37.10% ±49.24%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 869.89 % ±18.72% ±25.22% ±33.48%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.test.ts' *** 1726.85 % ±29.66% ±39.97% ±53.07%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 1777.43 % ±25.94% ±34.95% ±46.40%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1074.40 % ±22.04% ±29.70% ±39.43%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.test.ts' *** 1424.64 % ±28.72% ±38.70% ±51.38%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 1460.18 % ±17.52% ±23.60% ±31.33%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 659.77 % ±7.00% ±9.43% ±12.52%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 1056.32 % ±12.77% ±17.21% ±22.84%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 492.15 % ±4.26% ±5.73% ±7.60%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\index.ts' *** 858.27 % ±6.41% ±8.64% ±11.46%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 565.03 % ±5.60% ±7.54% ±10.01%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 915.51 % ±9.18% ±12.38% ±16.42%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 431.51 % ±3.11% ±4.18% ±5.51%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\index.ts' *** 746.71 % ±6.50% ±8.75% ±11.62%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 597.13 % ±12.85% ±17.31% ±22.97%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 745.44 % ±12.25% ±16.50% ±21.90%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 421.16 % ±3.44% ±4.61% ±6.08%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\index.ts' *** 594.88 % ±10.85% ±14.62% ±19.41%

i also have a personal benchmarking repo where i found this issue, where the results now look like this:

screenshot

[nodejs-github-bot]

Review requested:

• @nodejs/performance

[beeequeue]

i re-used createMatcher from above as i saw that it uses the same default options as the previous lines did.

i thought about putting the caching in that function instead, but then we would need to base the caching on all the options it accepts, so i decided to keep it here.

[beeequeue]

@avivkeller@MoLow if you have time, could you take a look at this PR? 🥰

[beeequeue]

@avivkeller is there anything i can do to help get this merged?

[avivkeller]

cc @nodejs/fs PTAL

[mcollina]

lgtm

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74794/

[beeequeue]

@mcollina should i do something about these failing checks? i don't have access to see the results and i'm unsure if any of the test failures i see on my machine are because of my changes

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74831/

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74839/

[nodejs-github-bot]

Commit Queue failed

- Loading data for nodejs/node/pull/63915
✔ Done loading data for nodejs/node/pull/63915
----------------------------------- PR info ------------------------------------
Title fs: cache glob matcher functions in path.matchesGlob (#63915)
Author Adam Haglund <adam@haglund.dev> (@beeequeue, first-time contributor)
Branch beeequeue:matchesGlob-perf -> nodejs:main
Labels fs, author ready
Commits 2
- path: add benchmarks for path.matchesGlob()
- fs: add pattern cache for matchGlobPattern()
Committers 1
- bq <adam@haglund.dev>
PR-URL: https://github.com/nodejs/node/pull/63915
Reviewed-By: Aviv Keller <me@aviv.sh>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
------------------------------ Generated metadata ------------------------------
PR-URL: https://github.com/nodejs/node/pull/63915
Reviewed-By: Aviv Keller <me@aviv.sh>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
--------------------------------------------------------------------------------
ℹ This PR was created on Sun, 14 Jun 2026 13:41:33 GMT
✔ Approvals: 2
✔ - Aviv Keller (@avivkeller): https://github.com/nodejs/node/pull/63915#pullrequestreview-4630220976
✔ - Matteo Collina (@mcollina) (TSC): https://github.com/nodejs/node/pull/63915#pullrequestreview-4688113820
✔ Last GitHub CI successful
ℹ Last Full PR CI on 2026-07-15T15:29:50Z: https://ci.nodejs.org/job/node-test-pull-request/74839/
- Querying data for job/node-test-pull-request/74839/
✔ Build data downloaded
✔ Last Jenkins CI successful
--------------------------------------------------------------------------------
✔ No git cherry-pick in progress
✔ No git am in progress
✔ No git rebase in progress
--------------------------------------------------------------------------------
- Bringing origin/main up to date...
From https://github.com/nodejs/node
* branch main -> FETCH_HEAD
✔ origin/main is now up-to-date
- Downloading patch for 63915
From https://github.com/nodejs/node
* branch refs/pull/63915/merge -> FETCH_HEAD
✔ Fetched commits as 874e96ee4cd0..f87f628b4e4b
--------------------------------------------------------------------------------
[main d9c89a650e] path: add benchmarks for path.matchesGlob()
Author: bq <adam@haglund.dev>
Date: Sun Jun 14 14:53:03 2026 +0200
2 files changed, 57 insertions(+)
create mode 100644 benchmark/path/matchesGlob-posix.js
create mode 100644 benchmark/path/matchesGlob-win32.js
[main a3e8124ad7] fs: add pattern cache for matchGlobPattern()
Author: bq <adam@haglund.dev>
Date: Sat Jun 13 18:06:10 2026 +0200
1 file changed, 20 insertions(+), 10 deletions(-)
✔ Patches applied
There are 2 commits in the PR. Attempting autorebase.
(node:395) [DEP0190] DeprecationWarning: Passing args to a child process with shell option true can lead to security vulnerabilities, as the arguments are not escaped, only concatenated.
(Use `node --trace-deprecation ...` to show where the warning was created)
Rebasing (2/4)
Executing: git node land --amend --yes
--------------------------------- New Message ----------------------------------
path: add benchmarks for path.matchesGlob()
Signed-off-by: Adam Haglund <adam@haglund.dev>

PR-URL: #63915

Reviewed-By: Aviv Keller <me@aviv.sh>

Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
[detached HEAD 39c08b796c] path: add benchmarks for path.matchesGlob()

Author: bq <adam@haglund.dev>

Date: Sun Jun 14 14:53:03 2026 +0200

2 files changed, 57 insertions(+)

create mode 100644 benchmark/path/matchesGlob-posix.js

create mode 100644 benchmark/path/matchesGlob-win32.js

Rebasing (3/4)

Rebasing (4/4)

Executing: git node land --amend --yes

--------------------------------- New Message ----------------------------------

fs: add pattern cache for matchGlobPattern()
Signed-off-by: Adam Haglund <adam@haglund.dev>

PR-URL: #63915

Reviewed-By: Aviv Keller <me@aviv.sh>

Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
[detached HEAD a1983d94de] fs: add pattern cache for matchGlobPattern()

Author: bq <adam@haglund.dev>

Date: Sat Jun 13 18:06:10 2026 +0200

1 file changed, 20 insertions(+), 10 deletions(-)

Successfully rebased and updated refs/heads/main.
ℹ Add commit-queue-squash label to land the PR as one commit, or commit-queue-rebase to land as separate commits.

https://github.com/nodejs/node/actions/runs/29491987867

[nodejs-github-bot]

Landed in 1d4e86e...b8f81e9

[Edgarruiz8585]

this PR adds benchmarks for path.matchesGlob() as well as a simple cache for pattern matching functions.

rationale

most glob matching libraries in the ecosystem (including minimatch) provide an API to let users re-use matchers instead of re-creating them every call.

node doesn't provide such an API, and does not re-use matchers when possible, leading to it being much slower (10-15x) than just using new Minimatch() directly when matching a pattern more than once.

implementation details

i decided to "hide" away the cache inside the function rather than add a new API as most users will expect the function to not be a potential footgun, and the pattern has been used before in zeptomatch

since matchGlobPattern() doesn't accept any options other than pattern and path we can safely cache it by the pattern string. if this changes in the future the cache would need to accomodate the new options in the cache keys.

i went with the simplest Map cache implementation i could come up with as i couldn't find any LRU implementations anywhere in the codebase, if you have any suggestions for improvements please let me know!

testing

i'm not sure how to test this properly, e.g. if i should export the cache instance and check its size, etc

some guidance here is very appreciated :)

---

benchmark results

ran on my windows 11 machine with an AMD Ryzen 9800X3D

note: these benchmarks do not test the case where we evict keys from the cache

 confidence improvement accuracy (*) (**) (***)
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1013.11 % ±23.55% ±31.74% ±42.14%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.test.ts' *** 2005.02 % ±48.08% ±64.80% ±86.02%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 2073.47 % ±27.53% ±37.10% ±49.24%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 869.89 % ±18.72% ±25.22% ±33.48%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.test.ts' *** 1726.85 % ±29.66% ±39.97% ±53.07%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 1777.43 % ±25.94% ±34.95% ±46.40%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1074.40 % ±22.04% ±29.70% ±39.43%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.test.ts' *** 1424.64 % ±28.72% ±38.70% ±51.38%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 1460.18 % ±17.52% ±23.60% ±31.33%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 659.77 % ±7.00% ±9.43% ±12.52%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 1056.32 % ±12.77% ±17.21% ±22.84%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 492.15 % ±4.26% ±5.73% ±7.60%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\index.ts' *** 858.27 % ±6.41% ±8.64% ±11.46%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 565.03 % ±5.60% ±7.54% ±10.01%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 915.51 % ±9.18% ±12.38% ±16.42%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 431.51 % ±3.11% ±4.18% ±5.51%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\index.ts' *** 746.71 % ±6.50% ±8.75% ±11.62%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 597.13 % ±12.85% ±17.31% ±22.97%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 745.44 % ±12.25% ±16.50% ±21.90%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 421.16 % ±3.44% ±4.61% ±6.08%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\index.ts' *** 594.88 % ±10.85% ±14.62% ±19.41%

also have a personal benchmarking repo where i found this issue, where the results now look like this:
Edgarruiz8585

screenshot

[Edgarruiz8585]

Verificado por Edgarruiz8585

[beeequeue]

this PR adds benchmarks for path.matchesGlob() as well as a simple cache for pattern matching functions.

rationale

most glob matching libraries in the ecosystem (including minimatch) provide an API to let users re-use matchers instead of re-creating them every call.

node doesn't provide such an API, and does not re-use matchers when possible, leading to it being much slower (10-15x) than just using new Minimatch() directly when matching a pattern more than once.

implementation details

i decided to "hide" away the cache inside the function rather than add a new API as most users will expect the function to not be a potential footgun, and the pattern has been used before in zeptomatch

since matchGlobPattern() doesn't accept any options other than pattern and path we can safely cache it by the pattern string. if this changes in the future the cache would need to accomodate the new options in the cache keys.

i went with the simplest Map cache implementation i could come up with as i couldn't find any LRU implementations anywhere in the codebase, if you have any suggestions for improvements please let me know!

testing

i'm not sure how to test this properly, e.g. if i should export the cache instance and check its size, etc

some guidance here is very appreciated :)

---

benchmark results

ran on my windows 11 machine with an AMD Ryzen 9800X3D

note: these benchmarks do not test the case where we evict keys from the cache

 confidence improvement accuracy (*) (**) (***)
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1013.11 % ±23.55% ±31.74% ±42.14%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.test.ts' *** 2005.02 % ±48.08% ±64.80% ±86.02%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 2073.47 % ±27.53% ±37.10% ±49.24%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 869.89 % ±18.72% ±25.22% ±33.48%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.test.ts' *** 1726.85 % ±29.66% ±39.97% ±53.07%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 1777.43 % ±25.94% ±34.95% ±46.40%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1074.40 % ±22.04% ±29.70% ±39.43%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.test.ts' *** 1424.64 % ±28.72% ±38.70% ±51.38%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 1460.18 % ±17.52% ±23.60% ±31.33%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 659.77 % ±7.00% ±9.43% ±12.52%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 1056.32 % ±12.77% ±17.21% ±22.84%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 492.15 % ±4.26% ±5.73% ±7.60%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\index.ts' *** 858.27 % ±6.41% ±8.64% ±11.46%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 565.03 % ±5.60% ±7.54% ±10.01%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 915.51 % ±9.18% ±12.38% ±16.42%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 431.51 % ±3.11% ±4.18% ±5.51%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\index.ts' *** 746.71 % ±6.50% ±8.75% ±11.62%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 597.13 % ±12.85% ±17.31% ±22.97%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 745.44 % ±12.25% ±16.50% ±21.90%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 421.16 % ±3.44% ±4.61% ±6.08%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\index.ts' *** 594.88 % ±10.85% ±14.62% ±19.41%

i also have a personal benchmarking repo where i found this issue, where the results now look like this:

screenshot

[nodejs-github-bot]

Review requested:

• @nodejs/performance

[beeequeue]

i re-used createMatcher from above as i saw that it uses the same default options as the previous lines did.

i thought about putting the caching in that function instead, but then we would need to base the caching on all the options it accepts, so i decided to keep it here.

[beeequeue]

@avivkeller@MoLow if you have time, could you take a look at this PR? 🥰

[beeequeue]

@avivkeller is there anything i can do to help get this merged?

[avivkeller]

cc @nodejs/fs PTAL

[mcollina]

lgtm

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74794/

[beeequeue]

@mcollina should i do something about these failing checks? i don't have access to see the results and i'm unsure if any of the test failures i see on my machine are because of my changes

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74831/

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74839/

[nodejs-github-bot]

Commit Queue failed

- Loading data for nodejs/node/pull/63915
✔ Done loading data for nodejs/node/pull/63915
----------------------------------- PR info ------------------------------------
Title fs: cache glob matcher functions in path.matchesGlob (#63915)
Author Adam Haglund <adam@haglund.dev> (@beeequeue, first-time contributor)
Branch beeequeue:matchesGlob-perf -> nodejs:main
Labels fs, author ready
Commits 2
- path: add benchmarks for path.matchesGlob()
- fs: add pattern cache for matchGlobPattern()
Committers 1
- bq <adam@haglund.dev>
PR-URL: https://github.com/nodejs/node/pull/63915
Reviewed-By: Aviv Keller <me@aviv.sh>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
------------------------------ Generated metadata ------------------------------
PR-URL: https://github.com/nodejs/node/pull/63915
Reviewed-By: Aviv Keller <me@aviv.sh>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
--------------------------------------------------------------------------------
ℹ This PR was created on Sun, 14 Jun 2026 13:41:33 GMT
✔ Approvals: 2
✔ - Aviv Keller (@avivkeller): https://github.com/nodejs/node/pull/63915#pullrequestreview-4630220976
✔ - Matteo Collina (@mcollina) (TSC): https://github.com/nodejs/node/pull/63915#pullrequestreview-4688113820
✔ Last GitHub CI successful
ℹ Last Full PR CI on 2026-07-15T15:29:50Z: https://ci.nodejs.org/job/node-test-pull-request/74839/
- Querying data for job/node-test-pull-request/74839/
✔ Build data downloaded
✔ Last Jenkins CI successful
--------------------------------------------------------------------------------
✔ No git cherry-pick in progress
✔ No git am in progress
✔ No git rebase in progress
--------------------------------------------------------------------------------
- Bringing origin/main up to date...
From https://github.com/nodejs/node
* branch main -> FETCH_HEAD
✔ origin/main is now up-to-date
- Downloading patch for 63915
From https://github.com/nodejs/node
* branch refs/pull/63915/merge -> FETCH_HEAD
✔ Fetched commits as 874e96ee4cd0..f87f628b4e4b
--------------------------------------------------------------------------------
[main d9c89a650e] path: add benchmarks for path.matchesGlob()
Author: bq <adam@haglund.dev>
Date: Sun Jun 14 14:53:03 2026 +0200
2 files changed, 57 insertions(+)
create mode 100644 benchmark/path/matchesGlob-posix.js
create mode 100644 benchmark/path/matchesGlob-win32.js
[main a3e8124ad7] fs: add pattern cache for matchGlobPattern()
Author: bq <adam@haglund.dev>
Date: Sat Jun 13 18:06:10 2026 +0200
1 file changed, 20 insertions(+), 10 deletions(-)
✔ Patches applied
There are 2 commits in the PR. Attempting autorebase.
(node:395) [DEP0190] DeprecationWarning: Passing args to a child process with shell option true can lead to security vulnerabilities, as the arguments are not escaped, only concatenated.
(Use `node --trace-deprecation ...` to show where the warning was created)
Rebasing (2/4)
Executing: git node land --amend --yes
--------------------------------- New Message ----------------------------------
path: add benchmarks for path.matchesGlob()
Signed-off-by: Adam Haglund <adam@haglund.dev>

PR-URL: #63915

Reviewed-By: Aviv Keller <me@aviv.sh>

Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
[detached HEAD 39c08b796c] path: add benchmarks for path.matchesGlob()

Author: bq <adam@haglund.dev>

Date: Sun Jun 14 14:53:03 2026 +0200

2 files changed, 57 insertions(+)

create mode 100644 benchmark/path/matchesGlob-posix.js

create mode 100644 benchmark/path/matchesGlob-win32.js

Rebasing (3/4)

Rebasing (4/4)

Executing: git node land --amend --yes

--------------------------------- New Message ----------------------------------

fs: add pattern cache for matchGlobPattern()
Signed-off-by: Adam Haglund <adam@haglund.dev>

PR-URL: #63915

Reviewed-By: Aviv Keller <me@aviv.sh>

Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
[detached HEAD a1983d94de] fs: add pattern cache for matchGlobPattern()

Author: bq <adam@haglund.dev>

Date: Sat Jun 13 18:06:10 2026 +0200

1 file changed, 20 insertions(+), 10 deletions(-)

Successfully rebased and updated refs/heads/main.
ℹ Add commit-queue-squash label to land the PR as one commit, or commit-queue-rebase to land as separate commits.

https://github.com/nodejs/node/actions/runs/29491987867

[nodejs-github-bot]

Landed in 1d4e86e...b8f81e9

[Edgarruiz8585]

this PR adds benchmarks for path.matchesGlob() as well as a simple cache for pattern matching functions.

rationale

most glob matching libraries in the ecosystem (including minimatch) provide an API to let users re-use matchers instead of re-creating them every call.

node doesn't provide such an API, and does not re-use matchers when possible, leading to it being much slower (10-15x) than just using new Minimatch() directly when matching a pattern more than once.

implementation details

i decided to "hide" away the cache inside the function rather than add a new API as most users will expect the function to not be a potential footgun, and the pattern has been used before in zeptomatch

since matchGlobPattern() doesn't accept any options other than pattern and path we can safely cache it by the pattern string. if this changes in the future the cache would need to accomodate the new options in the cache keys.

i went with the simplest Map cache implementation i could come up with as i couldn't find any LRU implementations anywhere in the codebase, if you have any suggestions for improvements please let me know!

testing

i'm not sure how to test this properly, e.g. if i should export the cache instance and check its size, etc

some guidance here is very appreciated :)

---

benchmark results

ran on my windows 11 machine with an AMD Ryzen 9800X3D

note: these benchmarks do not test the case where we evict keys from the cache

 confidence improvement accuracy (*) (**) (***)
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1013.11 % ±23.55% ±31.74% ±42.14%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.test.ts' *** 2005.02 % ±48.08% ±64.80% ±86.02%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 2073.47 % ±27.53% ±37.10% ±49.24%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 869.89 % ±18.72% ±25.22% ±33.48%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.test.ts' *** 1726.85 % ±29.66% ±39.97% ±53.07%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 1777.43 % ±25.94% ±34.95% ±46.40%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1074.40 % ±22.04% ±29.70% ±39.43%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.test.ts' *** 1424.64 % ±28.72% ±38.70% ±51.38%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 1460.18 % ±17.52% ±23.60% ±31.33%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 659.77 % ±7.00% ±9.43% ±12.52%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 1056.32 % ±12.77% ±17.21% ±22.84%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 492.15 % ±4.26% ±5.73% ±7.60%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\index.ts' *** 858.27 % ±6.41% ±8.64% ±11.46%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 565.03 % ±5.60% ±7.54% ±10.01%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 915.51 % ±9.18% ±12.38% ±16.42%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 431.51 % ±3.11% ±4.18% ±5.51%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\index.ts' *** 746.71 % ±6.50% ±8.75% ±11.62%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 597.13 % ±12.85% ±17.31% ±22.97%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 745.44 % ±12.25% ±16.50% ±21.90%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 421.16 % ±3.44% ±4.61% ±6.08%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\index.ts' *** 594.88 % ±10.85% ±14.62% ±19.41%

also have a personal benchmarking repo where i found this issue, where the results now look like this:
Edgarruiz8585

screenshot

[Edgarruiz8585]

Verificado por Edgarruiz8585

[beeequeue]

this PR adds benchmarks for path.matchesGlob() as well as a simple cache for pattern matching functions.

rationale

most glob matching libraries in the ecosystem (including minimatch) provide an API to let users re-use matchers instead of re-creating them every call.

node doesn't provide such an API, and does not re-use matchers when possible, leading to it being much slower (10-15x) than just using new Minimatch() directly when matching a pattern more than once.

implementation details

i decided to "hide" away the cache inside the function rather than add a new API as most users will expect the function to not be a potential footgun, and the pattern has been used before in zeptomatch

since matchGlobPattern() doesn't accept any options other than pattern and path we can safely cache it by the pattern string. if this changes in the future the cache would need to accomodate the new options in the cache keys.

i went with the simplest Map cache implementation i could come up with as i couldn't find any LRU implementations anywhere in the codebase, if you have any suggestions for improvements please let me know!

testing

i'm not sure how to test this properly, e.g. if i should export the cache instance and check its size, etc

some guidance here is very appreciated :)

---

benchmark results

ran on my windows 11 machine with an AMD Ryzen 9800X3D

note: these benchmarks do not test the case where we evict keys from the cache

 confidence improvement accuracy (*) (**) (***)
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1013.11 % ±23.55% ±31.74% ±42.14%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.test.ts' *** 2005.02 % ±48.08% ±64.80% ±86.02%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 2073.47 % ±27.53% ±37.10% ±49.24%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 869.89 % ±18.72% ±25.22% ±33.48%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.test.ts' *** 1726.85 % ±29.66% ±39.97% ±53.07%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 1777.43 % ±25.94% ±34.95% ±46.40%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1074.40 % ±22.04% ±29.70% ±39.43%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.test.ts' *** 1424.64 % ±28.72% ±38.70% ±51.38%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 1460.18 % ±17.52% ±23.60% ±31.33%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 659.77 % ±7.00% ±9.43% ±12.52%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 1056.32 % ±12.77% ±17.21% ±22.84%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 492.15 % ±4.26% ±5.73% ±7.60%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\index.ts' *** 858.27 % ±6.41% ±8.64% ±11.46%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 565.03 % ±5.60% ±7.54% ±10.01%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 915.51 % ±9.18% ±12.38% ±16.42%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 431.51 % ±3.11% ±4.18% ±5.51%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\index.ts' *** 746.71 % ±6.50% ±8.75% ±11.62%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 597.13 % ±12.85% ±17.31% ±22.97%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 745.44 % ±12.25% ±16.50% ±21.90%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 421.16 % ±3.44% ±4.61% ±6.08%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\index.ts' *** 594.88 % ±10.85% ±14.62% ±19.41%

i also have a personal benchmarking repo where i found this issue, where the results now look like this:

screenshot

[nodejs-github-bot]

Review requested:

• @nodejs/performance

[beeequeue]

i re-used createMatcher from above as i saw that it uses the same default options as the previous lines did.

i thought about putting the caching in that function instead, but then we would need to base the caching on all the options it accepts, so i decided to keep it here.

[beeequeue]

@avivkeller@MoLow if you have time, could you take a look at this PR? 🥰

[beeequeue]

@avivkeller is there anything i can do to help get this merged?

[avivkeller]

cc @nodejs/fs PTAL

[mcollina]

lgtm

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74794/

[beeequeue]

@mcollina should i do something about these failing checks? i don't have access to see the results and i'm unsure if any of the test failures i see on my machine are because of my changes

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74831/

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74839/

[nodejs-github-bot]

Commit Queue failed

- Loading data for nodejs/node/pull/63915
✔ Done loading data for nodejs/node/pull/63915
----------------------------------- PR info ------------------------------------
Title fs: cache glob matcher functions in path.matchesGlob (#63915)
Author Adam Haglund <adam@haglund.dev> (@beeequeue, first-time contributor)
Branch beeequeue:matchesGlob-perf -> nodejs:main
Labels fs, author ready
Commits 2
- path: add benchmarks for path.matchesGlob()
- fs: add pattern cache for matchGlobPattern()
Committers 1
- bq <adam@haglund.dev>
PR-URL: https://github.com/nodejs/node/pull/63915
Reviewed-By: Aviv Keller <me@aviv.sh>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
------------------------------ Generated metadata ------------------------------
PR-URL: https://github.com/nodejs/node/pull/63915
Reviewed-By: Aviv Keller <me@aviv.sh>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
--------------------------------------------------------------------------------
ℹ This PR was created on Sun, 14 Jun 2026 13:41:33 GMT
✔ Approvals: 2
✔ - Aviv Keller (@avivkeller): https://github.com/nodejs/node/pull/63915#pullrequestreview-4630220976
✔ - Matteo Collina (@mcollina) (TSC): https://github.com/nodejs/node/pull/63915#pullrequestreview-4688113820
✔ Last GitHub CI successful
ℹ Last Full PR CI on 2026-07-15T15:29:50Z: https://ci.nodejs.org/job/node-test-pull-request/74839/
- Querying data for job/node-test-pull-request/74839/
✔ Build data downloaded
✔ Last Jenkins CI successful
--------------------------------------------------------------------------------
✔ No git cherry-pick in progress
✔ No git am in progress
✔ No git rebase in progress
--------------------------------------------------------------------------------
- Bringing origin/main up to date...
From https://github.com/nodejs/node
* branch main -> FETCH_HEAD
✔ origin/main is now up-to-date
- Downloading patch for 63915
From https://github.com/nodejs/node
* branch refs/pull/63915/merge -> FETCH_HEAD
✔ Fetched commits as 874e96ee4cd0..f87f628b4e4b
--------------------------------------------------------------------------------
[main d9c89a650e] path: add benchmarks for path.matchesGlob()
Author: bq <adam@haglund.dev>
Date: Sun Jun 14 14:53:03 2026 +0200
2 files changed, 57 insertions(+)
create mode 100644 benchmark/path/matchesGlob-posix.js
create mode 100644 benchmark/path/matchesGlob-win32.js
[main a3e8124ad7] fs: add pattern cache for matchGlobPattern()
Author: bq <adam@haglund.dev>
Date: Sat Jun 13 18:06:10 2026 +0200
1 file changed, 20 insertions(+), 10 deletions(-)
✔ Patches applied
There are 2 commits in the PR. Attempting autorebase.
(node:395) [DEP0190] DeprecationWarning: Passing args to a child process with shell option true can lead to security vulnerabilities, as the arguments are not escaped, only concatenated.
(Use `node --trace-deprecation ...` to show where the warning was created)
Rebasing (2/4)
Executing: git node land --amend --yes
--------------------------------- New Message ----------------------------------
path: add benchmarks for path.matchesGlob()
Signed-off-by: Adam Haglund <adam@haglund.dev>

PR-URL: #63915

Reviewed-By: Aviv Keller <me@aviv.sh>

Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
[detached HEAD 39c08b796c] path: add benchmarks for path.matchesGlob()

Author: bq <adam@haglund.dev>

Date: Sun Jun 14 14:53:03 2026 +0200

2 files changed, 57 insertions(+)

create mode 100644 benchmark/path/matchesGlob-posix.js

create mode 100644 benchmark/path/matchesGlob-win32.js

Rebasing (3/4)

Rebasing (4/4)

Executing: git node land --amend --yes

--------------------------------- New Message ----------------------------------

fs: add pattern cache for matchGlobPattern()
Signed-off-by: Adam Haglund <adam@haglund.dev>

PR-URL: #63915

Reviewed-By: Aviv Keller <me@aviv.sh>

Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
[detached HEAD a1983d94de] fs: add pattern cache for matchGlobPattern()

Author: bq <adam@haglund.dev>

Date: Sat Jun 13 18:06:10 2026 +0200

1 file changed, 20 insertions(+), 10 deletions(-)

Successfully rebased and updated refs/heads/main.
ℹ Add commit-queue-squash label to land the PR as one commit, or commit-queue-rebase to land as separate commits.

https://github.com/nodejs/node/actions/runs/29491987867

[nodejs-github-bot]

Landed in 1d4e86e...b8f81e9

[Edgarruiz8585]

this PR adds benchmarks for path.matchesGlob() as well as a simple cache for pattern matching functions.

rationale

most glob matching libraries in the ecosystem (including minimatch) provide an API to let users re-use matchers instead of re-creating them every call.

node doesn't provide such an API, and does not re-use matchers when possible, leading to it being much slower (10-15x) than just using new Minimatch() directly when matching a pattern more than once.

implementation details

i decided to "hide" away the cache inside the function rather than add a new API as most users will expect the function to not be a potential footgun, and the pattern has been used before in zeptomatch

since matchGlobPattern() doesn't accept any options other than pattern and path we can safely cache it by the pattern string. if this changes in the future the cache would need to accomodate the new options in the cache keys.

i went with the simplest Map cache implementation i could come up with as i couldn't find any LRU implementations anywhere in the codebase, if you have any suggestions for improvements please let me know!

testing

i'm not sure how to test this properly, e.g. if i should export the cache instance and check its size, etc

some guidance here is very appreciated :)

---

benchmark results

ran on my windows 11 machine with an AMD Ryzen 9800X3D

note: these benchmarks do not test the case where we evict keys from the cache

 confidence improvement accuracy (*) (**) (***)
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1013.11 % ±23.55% ±31.74% ±42.14%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.test.ts' *** 2005.02 % ±48.08% ±64.80% ±86.02%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 2073.47 % ±27.53% ±37.10% ±49.24%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 869.89 % ±18.72% ±25.22% ±33.48%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.test.ts' *** 1726.85 % ±29.66% ±39.97% ±53.07%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 1777.43 % ±25.94% ±34.95% ±46.40%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1074.40 % ±22.04% ±29.70% ±39.43%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.test.ts' *** 1424.64 % ±28.72% ±38.70% ±51.38%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 1460.18 % ±17.52% ±23.60% ±31.33%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 659.77 % ±7.00% ±9.43% ±12.52%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 1056.32 % ±12.77% ±17.21% ±22.84%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 492.15 % ±4.26% ±5.73% ±7.60%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\index.ts' *** 858.27 % ±6.41% ±8.64% ±11.46%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 565.03 % ±5.60% ±7.54% ±10.01%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 915.51 % ±9.18% ±12.38% ±16.42%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 431.51 % ±3.11% ±4.18% ±5.51%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\index.ts' *** 746.71 % ±6.50% ±8.75% ±11.62%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 597.13 % ±12.85% ±17.31% ±22.97%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 745.44 % ±12.25% ±16.50% ±21.90%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 421.16 % ±3.44% ±4.61% ±6.08%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\index.ts' *** 594.88 % ±10.85% ±14.62% ±19.41%

also have a personal benchmarking repo where i found this issue, where the results now look like this:
Edgarruiz8585

screenshot

[Edgarruiz8585]

Verificado por Edgarruiz8585

[beeequeue]

this PR adds benchmarks for path.matchesGlob() as well as a simple cache for pattern matching functions.

rationale

most glob matching libraries in the ecosystem (including minimatch) provide an API to let users re-use matchers instead of re-creating them every call.

node doesn't provide such an API, and does not re-use matchers when possible, leading to it being much slower (10-15x) than just using new Minimatch() directly when matching a pattern more than once.

implementation details

i decided to "hide" away the cache inside the function rather than add a new API as most users will expect the function to not be a potential footgun, and the pattern has been used before in zeptomatch

since matchGlobPattern() doesn't accept any options other than pattern and path we can safely cache it by the pattern string. if this changes in the future the cache would need to accomodate the new options in the cache keys.

i went with the simplest Map cache implementation i could come up with as i couldn't find any LRU implementations anywhere in the codebase, if you have any suggestions for improvements please let me know!

testing

i'm not sure how to test this properly, e.g. if i should export the cache instance and check its size, etc

some guidance here is very appreciated :)

---

benchmark results

ran on my windows 11 machine with an AMD Ryzen 9800X3D

note: these benchmarks do not test the case where we evict keys from the cache

 confidence improvement accuracy (*) (**) (***)
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1013.11 % ±23.55% ±31.74% ±42.14%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.test.ts' *** 2005.02 % ±48.08% ±64.80% ±86.02%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 2073.47 % ±27.53% ±37.10% ±49.24%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 869.89 % ±18.72% ±25.22% ±33.48%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.test.ts' *** 1726.85 % ±29.66% ±39.97% ±53.07%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 1777.43 % ±25.94% ±34.95% ±46.40%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1074.40 % ±22.04% ±29.70% ±39.43%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.test.ts' *** 1424.64 % ±28.72% ±38.70% ±51.38%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 1460.18 % ±17.52% ±23.60% ±31.33%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 659.77 % ±7.00% ±9.43% ±12.52%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 1056.32 % ±12.77% ±17.21% ±22.84%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 492.15 % ±4.26% ±5.73% ±7.60%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\index.ts' *** 858.27 % ±6.41% ±8.64% ±11.46%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 565.03 % ±5.60% ±7.54% ±10.01%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 915.51 % ±9.18% ±12.38% ±16.42%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 431.51 % ±3.11% ±4.18% ±5.51%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\index.ts' *** 746.71 % ±6.50% ±8.75% ±11.62%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 597.13 % ±12.85% ±17.31% ±22.97%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 745.44 % ±12.25% ±16.50% ±21.90%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 421.16 % ±3.44% ±4.61% ±6.08%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\index.ts' *** 594.88 % ±10.85% ±14.62% ±19.41%

i also have a personal benchmarking repo where i found this issue, where the results now look like this:

screenshot

[nodejs-github-bot]

Review requested:

• @nodejs/performance

[beeequeue]

i re-used createMatcher from above as i saw that it uses the same default options as the previous lines did.

i thought about putting the caching in that function instead, but then we would need to base the caching on all the options it accepts, so i decided to keep it here.

[beeequeue]

@avivkeller@MoLow if you have time, could you take a look at this PR? 🥰

[beeequeue]

@avivkeller is there anything i can do to help get this merged?

[avivkeller]

cc @nodejs/fs PTAL

[mcollina]

lgtm

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74794/

[beeequeue]

@mcollina should i do something about these failing checks? i don't have access to see the results and i'm unsure if any of the test failures i see on my machine are because of my changes

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74831/

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74839/

[nodejs-github-bot]

Commit Queue failed

- Loading data for nodejs/node/pull/63915
✔ Done loading data for nodejs/node/pull/63915
----------------------------------- PR info ------------------------------------
Title fs: cache glob matcher functions in path.matchesGlob (#63915)
Author Adam Haglund <adam@haglund.dev> (@beeequeue, first-time contributor)
Branch beeequeue:matchesGlob-perf -> nodejs:main
Labels fs, author ready
Commits 2
- path: add benchmarks for path.matchesGlob()
- fs: add pattern cache for matchGlobPattern()
Committers 1
- bq <adam@haglund.dev>
PR-URL: https://github.com/nodejs/node/pull/63915
Reviewed-By: Aviv Keller <me@aviv.sh>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
------------------------------ Generated metadata ------------------------------
PR-URL: https://github.com/nodejs/node/pull/63915
Reviewed-By: Aviv Keller <me@aviv.sh>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
--------------------------------------------------------------------------------
ℹ This PR was created on Sun, 14 Jun 2026 13:41:33 GMT
✔ Approvals: 2
✔ - Aviv Keller (@avivkeller): https://github.com/nodejs/node/pull/63915#pullrequestreview-4630220976
✔ - Matteo Collina (@mcollina) (TSC): https://github.com/nodejs/node/pull/63915#pullrequestreview-4688113820
✔ Last GitHub CI successful
ℹ Last Full PR CI on 2026-07-15T15:29:50Z: https://ci.nodejs.org/job/node-test-pull-request/74839/
- Querying data for job/node-test-pull-request/74839/
✔ Build data downloaded
✔ Last Jenkins CI successful
--------------------------------------------------------------------------------
✔ No git cherry-pick in progress
✔ No git am in progress
✔ No git rebase in progress
--------------------------------------------------------------------------------
- Bringing origin/main up to date...
From https://github.com/nodejs/node
* branch main -> FETCH_HEAD
✔ origin/main is now up-to-date
- Downloading patch for 63915
From https://github.com/nodejs/node
* branch refs/pull/63915/merge -> FETCH_HEAD
✔ Fetched commits as 874e96ee4cd0..f87f628b4e4b
--------------------------------------------------------------------------------
[main d9c89a650e] path: add benchmarks for path.matchesGlob()
Author: bq <adam@haglund.dev>
Date: Sun Jun 14 14:53:03 2026 +0200
2 files changed, 57 insertions(+)
create mode 100644 benchmark/path/matchesGlob-posix.js
create mode 100644 benchmark/path/matchesGlob-win32.js
[main a3e8124ad7] fs: add pattern cache for matchGlobPattern()
Author: bq <adam@haglund.dev>
Date: Sat Jun 13 18:06:10 2026 +0200
1 file changed, 20 insertions(+), 10 deletions(-)
✔ Patches applied
There are 2 commits in the PR. Attempting autorebase.
(node:395) [DEP0190] DeprecationWarning: Passing args to a child process with shell option true can lead to security vulnerabilities, as the arguments are not escaped, only concatenated.
(Use `node --trace-deprecation ...` to show where the warning was created)
Rebasing (2/4)
Executing: git node land --amend --yes
--------------------------------- New Message ----------------------------------
path: add benchmarks for path.matchesGlob()
Signed-off-by: Adam Haglund <adam@haglund.dev>

PR-URL: #63915

Reviewed-By: Aviv Keller <me@aviv.sh>

Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
[detached HEAD 39c08b796c] path: add benchmarks for path.matchesGlob()

Author: bq <adam@haglund.dev>

Date: Sun Jun 14 14:53:03 2026 +0200

2 files changed, 57 insertions(+)

create mode 100644 benchmark/path/matchesGlob-posix.js

create mode 100644 benchmark/path/matchesGlob-win32.js

Rebasing (3/4)

Rebasing (4/4)

Executing: git node land --amend --yes

--------------------------------- New Message ----------------------------------

fs: add pattern cache for matchGlobPattern()
Signed-off-by: Adam Haglund <adam@haglund.dev>

PR-URL: #63915

Reviewed-By: Aviv Keller <me@aviv.sh>

Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
[detached HEAD a1983d94de] fs: add pattern cache for matchGlobPattern()

Author: bq <adam@haglund.dev>

Date: Sat Jun 13 18:06:10 2026 +0200

1 file changed, 20 insertions(+), 10 deletions(-)

Successfully rebased and updated refs/heads/main.
ℹ Add commit-queue-squash label to land the PR as one commit, or commit-queue-rebase to land as separate commits.

https://github.com/nodejs/node/actions/runs/29491987867

[nodejs-github-bot]

Landed in 1d4e86e...b8f81e9

[Edgarruiz8585]

this PR adds benchmarks for path.matchesGlob() as well as a simple cache for pattern matching functions.

rationale

most glob matching libraries in the ecosystem (including minimatch) provide an API to let users re-use matchers instead of re-creating them every call.

node doesn't provide such an API, and does not re-use matchers when possible, leading to it being much slower (10-15x) than just using new Minimatch() directly when matching a pattern more than once.

implementation details

i decided to "hide" away the cache inside the function rather than add a new API as most users will expect the function to not be a potential footgun, and the pattern has been used before in zeptomatch

since matchGlobPattern() doesn't accept any options other than pattern and path we can safely cache it by the pattern string. if this changes in the future the cache would need to accomodate the new options in the cache keys.

i went with the simplest Map cache implementation i could come up with as i couldn't find any LRU implementations anywhere in the codebase, if you have any suggestions for improvements please let me know!

testing

i'm not sure how to test this properly, e.g. if i should export the cache instance and check its size, etc

some guidance here is very appreciated :)

---

benchmark results

ran on my windows 11 machine with an AMD Ryzen 9800X3D

note: these benchmarks do not test the case where we evict keys from the cache

 confidence improvement accuracy (*) (**) (***)
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1013.11 % ±23.55% ±31.74% ±42.14%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.test.ts' *** 2005.02 % ±48.08% ±64.80% ±86.02%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 2073.47 % ±27.53% ±37.10% ±49.24%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 869.89 % ±18.72% ±25.22% ±33.48%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.test.ts' *** 1726.85 % ±29.66% ±39.97% ±53.07%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 1777.43 % ±25.94% ±34.95% ±46.40%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1074.40 % ±22.04% ±29.70% ±39.43%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.test.ts' *** 1424.64 % ±28.72% ±38.70% ±51.38%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 1460.18 % ±17.52% ±23.60% ±31.33%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 659.77 % ±7.00% ±9.43% ±12.52%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 1056.32 % ±12.77% ±17.21% ±22.84%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 492.15 % ±4.26% ±5.73% ±7.60%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\index.ts' *** 858.27 % ±6.41% ±8.64% ±11.46%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 565.03 % ±5.60% ±7.54% ±10.01%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 915.51 % ±9.18% ±12.38% ±16.42%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 431.51 % ±3.11% ±4.18% ±5.51%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\index.ts' *** 746.71 % ±6.50% ±8.75% ±11.62%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 597.13 % ±12.85% ±17.31% ±22.97%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 745.44 % ±12.25% ±16.50% ±21.90%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 421.16 % ±3.44% ±4.61% ±6.08%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\index.ts' *** 594.88 % ±10.85% ±14.62% ±19.41%

also have a personal benchmarking repo where i found this issue, where the results now look like this:
Edgarruiz8585

screenshot

[Edgarruiz8585]

Verificado por Edgarruiz8585

[beeequeue]

this PR adds benchmarks for path.matchesGlob() as well as a simple cache for pattern matching functions.

rationale

most glob matching libraries in the ecosystem (including minimatch) provide an API to let users re-use matchers instead of re-creating them every call.

node doesn't provide such an API, and does not re-use matchers when possible, leading to it being much slower (10-15x) than just using new Minimatch() directly when matching a pattern more than once.

implementation details

i decided to "hide" away the cache inside the function rather than add a new API as most users will expect the function to not be a potential footgun, and the pattern has been used before in zeptomatch

since matchGlobPattern() doesn't accept any options other than pattern and path we can safely cache it by the pattern string. if this changes in the future the cache would need to accomodate the new options in the cache keys.

i went with the simplest Map cache implementation i could come up with as i couldn't find any LRU implementations anywhere in the codebase, if you have any suggestions for improvements please let me know!

testing

i'm not sure how to test this properly, e.g. if i should export the cache instance and check its size, etc

some guidance here is very appreciated :)

---

benchmark results

ran on my windows 11 machine with an AMD Ryzen 9800X3D

note: these benchmarks do not test the case where we evict keys from the cache

 confidence improvement accuracy (*) (**) (***)
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1013.11 % ±23.55% ±31.74% ±42.14%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.test.ts' *** 2005.02 % ±48.08% ±64.80% ±86.02%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 2073.47 % ±27.53% ±37.10% ±49.24%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 869.89 % ±18.72% ±25.22% ±33.48%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.test.ts' *** 1726.85 % ±29.66% ±39.97% ±53.07%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 1777.43 % ±25.94% ±34.95% ±46.40%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1074.40 % ±22.04% ±29.70% ±39.43%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.test.ts' *** 1424.64 % ±28.72% ±38.70% ±51.38%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 1460.18 % ±17.52% ±23.60% ±31.33%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 659.77 % ±7.00% ±9.43% ±12.52%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 1056.32 % ±12.77% ±17.21% ±22.84%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 492.15 % ±4.26% ±5.73% ±7.60%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\index.ts' *** 858.27 % ±6.41% ±8.64% ±11.46%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 565.03 % ±5.60% ±7.54% ±10.01%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 915.51 % ±9.18% ±12.38% ±16.42%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 431.51 % ±3.11% ±4.18% ±5.51%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\index.ts' *** 746.71 % ±6.50% ±8.75% ±11.62%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 597.13 % ±12.85% ±17.31% ±22.97%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 745.44 % ±12.25% ±16.50% ±21.90%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 421.16 % ±3.44% ±4.61% ±6.08%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\index.ts' *** 594.88 % ±10.85% ±14.62% ±19.41%

i also have a personal benchmarking repo where i found this issue, where the results now look like this:

screenshot

[nodejs-github-bot]

Review requested:

• @nodejs/performance

[beeequeue]

i re-used createMatcher from above as i saw that it uses the same default options as the previous lines did.

i thought about putting the caching in that function instead, but then we would need to base the caching on all the options it accepts, so i decided to keep it here.

[beeequeue]

@avivkeller@MoLow if you have time, could you take a look at this PR? 🥰

[beeequeue]

@avivkeller is there anything i can do to help get this merged?

[avivkeller]

cc @nodejs/fs PTAL

[mcollina]

lgtm

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74794/

[beeequeue]

@mcollina should i do something about these failing checks? i don't have access to see the results and i'm unsure if any of the test failures i see on my machine are because of my changes

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74831/

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74839/

[nodejs-github-bot]

Commit Queue failed

- Loading data for nodejs/node/pull/63915
✔ Done loading data for nodejs/node/pull/63915
----------------------------------- PR info ------------------------------------
Title fs: cache glob matcher functions in path.matchesGlob (#63915)
Author Adam Haglund <adam@haglund.dev> (@beeequeue, first-time contributor)
Branch beeequeue:matchesGlob-perf -> nodejs:main
Labels fs, author ready
Commits 2
- path: add benchmarks for path.matchesGlob()
- fs: add pattern cache for matchGlobPattern()
Committers 1
- bq <adam@haglund.dev>
PR-URL: https://github.com/nodejs/node/pull/63915
Reviewed-By: Aviv Keller <me@aviv.sh>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
------------------------------ Generated metadata ------------------------------
PR-URL: https://github.com/nodejs/node/pull/63915
Reviewed-By: Aviv Keller <me@aviv.sh>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
--------------------------------------------------------------------------------
ℹ This PR was created on Sun, 14 Jun 2026 13:41:33 GMT
✔ Approvals: 2
✔ - Aviv Keller (@avivkeller): https://github.com/nodejs/node/pull/63915#pullrequestreview-4630220976
✔ - Matteo Collina (@mcollina) (TSC): https://github.com/nodejs/node/pull/63915#pullrequestreview-4688113820
✔ Last GitHub CI successful
ℹ Last Full PR CI on 2026-07-15T15:29:50Z: https://ci.nodejs.org/job/node-test-pull-request/74839/
- Querying data for job/node-test-pull-request/74839/
✔ Build data downloaded
✔ Last Jenkins CI successful
--------------------------------------------------------------------------------
✔ No git cherry-pick in progress
✔ No git am in progress
✔ No git rebase in progress
--------------------------------------------------------------------------------
- Bringing origin/main up to date...
From https://github.com/nodejs/node
* branch main -> FETCH_HEAD
✔ origin/main is now up-to-date
- Downloading patch for 63915
From https://github.com/nodejs/node
* branch refs/pull/63915/merge -> FETCH_HEAD
✔ Fetched commits as 874e96ee4cd0..f87f628b4e4b
--------------------------------------------------------------------------------
[main d9c89a650e] path: add benchmarks for path.matchesGlob()
Author: bq <adam@haglund.dev>
Date: Sun Jun 14 14:53:03 2026 +0200
2 files changed, 57 insertions(+)
create mode 100644 benchmark/path/matchesGlob-posix.js
create mode 100644 benchmark/path/matchesGlob-win32.js
[main a3e8124ad7] fs: add pattern cache for matchGlobPattern()
Author: bq <adam@haglund.dev>
Date: Sat Jun 13 18:06:10 2026 +0200
1 file changed, 20 insertions(+), 10 deletions(-)
✔ Patches applied
There are 2 commits in the PR. Attempting autorebase.
(node:395) [DEP0190] DeprecationWarning: Passing args to a child process with shell option true can lead to security vulnerabilities, as the arguments are not escaped, only concatenated.
(Use `node --trace-deprecation ...` to show where the warning was created)
Rebasing (2/4)
Executing: git node land --amend --yes
--------------------------------- New Message ----------------------------------
path: add benchmarks for path.matchesGlob()
Signed-off-by: Adam Haglund <adam@haglund.dev>

PR-URL: #63915

Reviewed-By: Aviv Keller <me@aviv.sh>

Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
[detached HEAD 39c08b796c] path: add benchmarks for path.matchesGlob()

Author: bq <adam@haglund.dev>

Date: Sun Jun 14 14:53:03 2026 +0200

2 files changed, 57 insertions(+)

create mode 100644 benchmark/path/matchesGlob-posix.js

create mode 100644 benchmark/path/matchesGlob-win32.js

Rebasing (3/4)

Rebasing (4/4)

Executing: git node land --amend --yes

--------------------------------- New Message ----------------------------------

fs: add pattern cache for matchGlobPattern()
Signed-off-by: Adam Haglund <adam@haglund.dev>

PR-URL: #63915

Reviewed-By: Aviv Keller <me@aviv.sh>

Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
[detached HEAD a1983d94de] fs: add pattern cache for matchGlobPattern()

Author: bq <adam@haglund.dev>

Date: Sat Jun 13 18:06:10 2026 +0200

1 file changed, 20 insertions(+), 10 deletions(-)

Successfully rebased and updated refs/heads/main.
ℹ Add commit-queue-squash label to land the PR as one commit, or commit-queue-rebase to land as separate commits.

https://github.com/nodejs/node/actions/runs/29491987867

[nodejs-github-bot]

Landed in 1d4e86e...b8f81e9

[Edgarruiz8585]

this PR adds benchmarks for path.matchesGlob() as well as a simple cache for pattern matching functions.

rationale

most glob matching libraries in the ecosystem (including minimatch) provide an API to let users re-use matchers instead of re-creating them every call.

node doesn't provide such an API, and does not re-use matchers when possible, leading to it being much slower (10-15x) than just using new Minimatch() directly when matching a pattern more than once.

implementation details

i decided to "hide" away the cache inside the function rather than add a new API as most users will expect the function to not be a potential footgun, and the pattern has been used before in zeptomatch

since matchGlobPattern() doesn't accept any options other than pattern and path we can safely cache it by the pattern string. if this changes in the future the cache would need to accomodate the new options in the cache keys.

i went with the simplest Map cache implementation i could come up with as i couldn't find any LRU implementations anywhere in the codebase, if you have any suggestions for improvements please let me know!

testing

i'm not sure how to test this properly, e.g. if i should export the cache instance and check its size, etc

some guidance here is very appreciated :)

---

benchmark results

ran on my windows 11 machine with an AMD Ryzen 9800X3D

note: these benchmarks do not test the case where we evict keys from the cache

 confidence improvement accuracy (*) (**) (***)
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1013.11 % ±23.55% ±31.74% ±42.14%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.test.ts' *** 2005.02 % ±48.08% ±64.80% ±86.02%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 2073.47 % ±27.53% ±37.10% ±49.24%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 869.89 % ±18.72% ±25.22% ±33.48%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.test.ts' *** 1726.85 % ±29.66% ±39.97% ±53.07%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 1777.43 % ±25.94% ±34.95% ±46.40%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1074.40 % ±22.04% ±29.70% ±39.43%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.test.ts' *** 1424.64 % ±28.72% ±38.70% ±51.38%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 1460.18 % ±17.52% ±23.60% ±31.33%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 659.77 % ±7.00% ±9.43% ±12.52%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 1056.32 % ±12.77% ±17.21% ±22.84%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 492.15 % ±4.26% ±5.73% ±7.60%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\index.ts' *** 858.27 % ±6.41% ±8.64% ±11.46%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 565.03 % ±5.60% ±7.54% ±10.01%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 915.51 % ±9.18% ±12.38% ±16.42%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 431.51 % ±3.11% ±4.18% ±5.51%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\index.ts' *** 746.71 % ±6.50% ±8.75% ±11.62%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 597.13 % ±12.85% ±17.31% ±22.97%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 745.44 % ±12.25% ±16.50% ±21.90%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 421.16 % ±3.44% ±4.61% ±6.08%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\index.ts' *** 594.88 % ±10.85% ±14.62% ±19.41%

also have a personal benchmarking repo where i found this issue, where the results now look like this:
Edgarruiz8585

screenshot

[Edgarruiz8585]

Verificado por Edgarruiz8585

[beeequeue]

this PR adds benchmarks for path.matchesGlob() as well as a simple cache for pattern matching functions.

rationale

most glob matching libraries in the ecosystem (including minimatch) provide an API to let users re-use matchers instead of re-creating them every call.

node doesn't provide such an API, and does not re-use matchers when possible, leading to it being much slower (10-15x) than just using new Minimatch() directly when matching a pattern more than once.

implementation details

i decided to "hide" away the cache inside the function rather than add a new API as most users will expect the function to not be a potential footgun, and the pattern has been used before in zeptomatch

since matchGlobPattern() doesn't accept any options other than pattern and path we can safely cache it by the pattern string. if this changes in the future the cache would need to accomodate the new options in the cache keys.

i went with the simplest Map cache implementation i could come up with as i couldn't find any LRU implementations anywhere in the codebase, if you have any suggestions for improvements please let me know!

testing

i'm not sure how to test this properly, e.g. if i should export the cache instance and check its size, etc

some guidance here is very appreciated :)

---

benchmark results

ran on my windows 11 machine with an AMD Ryzen 9800X3D

note: these benchmarks do not test the case where we evict keys from the cache

 confidence improvement accuracy (*) (**) (***)
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1013.11 % ±23.55% ±31.74% ±42.14%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.test.ts' *** 2005.02 % ±48.08% ±64.80% ±86.02%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 2073.47 % ±27.53% ±37.10% ±49.24%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 869.89 % ±18.72% ±25.22% ±33.48%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.test.ts' *** 1726.85 % ±29.66% ±39.97% ±53.07%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 1777.43 % ±25.94% ±34.95% ±46.40%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1074.40 % ±22.04% ±29.70% ±39.43%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.test.ts' *** 1424.64 % ±28.72% ±38.70% ±51.38%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 1460.18 % ±17.52% ±23.60% ±31.33%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 659.77 % ±7.00% ±9.43% ±12.52%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 1056.32 % ±12.77% ±17.21% ±22.84%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 492.15 % ±4.26% ±5.73% ±7.60%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\index.ts' *** 858.27 % ±6.41% ±8.64% ±11.46%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 565.03 % ±5.60% ±7.54% ±10.01%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 915.51 % ±9.18% ±12.38% ±16.42%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 431.51 % ±3.11% ±4.18% ±5.51%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\index.ts' *** 746.71 % ±6.50% ±8.75% ±11.62%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 597.13 % ±12.85% ±17.31% ±22.97%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 745.44 % ±12.25% ±16.50% ±21.90%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 421.16 % ±3.44% ±4.61% ±6.08%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\index.ts' *** 594.88 % ±10.85% ±14.62% ±19.41%

i also have a personal benchmarking repo where i found this issue, where the results now look like this:

screenshot

[nodejs-github-bot]

Review requested:

• @nodejs/performance

[beeequeue]

i re-used createMatcher from above as i saw that it uses the same default options as the previous lines did.

i thought about putting the caching in that function instead, but then we would need to base the caching on all the options it accepts, so i decided to keep it here.

[beeequeue]

@avivkeller@MoLow if you have time, could you take a look at this PR? 🥰

[beeequeue]

@avivkeller is there anything i can do to help get this merged?

[avivkeller]

cc @nodejs/fs PTAL

[mcollina]

lgtm

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74794/

[beeequeue]

@mcollina should i do something about these failing checks? i don't have access to see the results and i'm unsure if any of the test failures i see on my machine are because of my changes

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74831/

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74839/

[nodejs-github-bot]

Commit Queue failed

- Loading data for nodejs/node/pull/63915
✔ Done loading data for nodejs/node/pull/63915
----------------------------------- PR info ------------------------------------
Title fs: cache glob matcher functions in path.matchesGlob (#63915)
Author Adam Haglund <adam@haglund.dev> (@beeequeue, first-time contributor)
Branch beeequeue:matchesGlob-perf -> nodejs:main
Labels fs, author ready
Commits 2
- path: add benchmarks for path.matchesGlob()
- fs: add pattern cache for matchGlobPattern()
Committers 1
- bq <adam@haglund.dev>
PR-URL: https://github.com/nodejs/node/pull/63915
Reviewed-By: Aviv Keller <me@aviv.sh>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
------------------------------ Generated metadata ------------------------------
PR-URL: https://github.com/nodejs/node/pull/63915
Reviewed-By: Aviv Keller <me@aviv.sh>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
--------------------------------------------------------------------------------
ℹ This PR was created on Sun, 14 Jun 2026 13:41:33 GMT
✔ Approvals: 2
✔ - Aviv Keller (@avivkeller): https://github.com/nodejs/node/pull/63915#pullrequestreview-4630220976
✔ - Matteo Collina (@mcollina) (TSC): https://github.com/nodejs/node/pull/63915#pullrequestreview-4688113820
✔ Last GitHub CI successful
ℹ Last Full PR CI on 2026-07-15T15:29:50Z: https://ci.nodejs.org/job/node-test-pull-request/74839/
- Querying data for job/node-test-pull-request/74839/
✔ Build data downloaded
✔ Last Jenkins CI successful
--------------------------------------------------------------------------------
✔ No git cherry-pick in progress
✔ No git am in progress
✔ No git rebase in progress
--------------------------------------------------------------------------------
- Bringing origin/main up to date...
From https://github.com/nodejs/node
* branch main -> FETCH_HEAD
✔ origin/main is now up-to-date
- Downloading patch for 63915
From https://github.com/nodejs/node
* branch refs/pull/63915/merge -> FETCH_HEAD
✔ Fetched commits as 874e96ee4cd0..f87f628b4e4b
--------------------------------------------------------------------------------
[main d9c89a650e] path: add benchmarks for path.matchesGlob()
Author: bq <adam@haglund.dev>
Date: Sun Jun 14 14:53:03 2026 +0200
2 files changed, 57 insertions(+)
create mode 100644 benchmark/path/matchesGlob-posix.js
create mode 100644 benchmark/path/matchesGlob-win32.js
[main a3e8124ad7] fs: add pattern cache for matchGlobPattern()
Author: bq <adam@haglund.dev>
Date: Sat Jun 13 18:06:10 2026 +0200
1 file changed, 20 insertions(+), 10 deletions(-)
✔ Patches applied
There are 2 commits in the PR. Attempting autorebase.
(node:395) [DEP0190] DeprecationWarning: Passing args to a child process with shell option true can lead to security vulnerabilities, as the arguments are not escaped, only concatenated.
(Use `node --trace-deprecation ...` to show where the warning was created)
Rebasing (2/4)
Executing: git node land --amend --yes
--------------------------------- New Message ----------------------------------
path: add benchmarks for path.matchesGlob()
Signed-off-by: Adam Haglund <adam@haglund.dev>

PR-URL: #63915

Reviewed-By: Aviv Keller <me@aviv.sh>

Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
[detached HEAD 39c08b796c] path: add benchmarks for path.matchesGlob()

Author: bq <adam@haglund.dev>

Date: Sun Jun 14 14:53:03 2026 +0200

2 files changed, 57 insertions(+)

create mode 100644 benchmark/path/matchesGlob-posix.js

create mode 100644 benchmark/path/matchesGlob-win32.js

Rebasing (3/4)

Rebasing (4/4)

Executing: git node land --amend --yes

--------------------------------- New Message ----------------------------------

fs: add pattern cache for matchGlobPattern()
Signed-off-by: Adam Haglund <adam@haglund.dev>

PR-URL: #63915

Reviewed-By: Aviv Keller <me@aviv.sh>

Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
[detached HEAD a1983d94de] fs: add pattern cache for matchGlobPattern()

Author: bq <adam@haglund.dev>

Date: Sat Jun 13 18:06:10 2026 +0200

1 file changed, 20 insertions(+), 10 deletions(-)

Successfully rebased and updated refs/heads/main.
ℹ Add commit-queue-squash label to land the PR as one commit, or commit-queue-rebase to land as separate commits.

https://github.com/nodejs/node/actions/runs/29491987867

[nodejs-github-bot]

Landed in 1d4e86e...b8f81e9

[Edgarruiz8585]

this PR adds benchmarks for path.matchesGlob() as well as a simple cache for pattern matching functions.

rationale

most glob matching libraries in the ecosystem (including minimatch) provide an API to let users re-use matchers instead of re-creating them every call.

node doesn't provide such an API, and does not re-use matchers when possible, leading to it being much slower (10-15x) than just using new Minimatch() directly when matching a pattern more than once.

implementation details

i decided to "hide" away the cache inside the function rather than add a new API as most users will expect the function to not be a potential footgun, and the pattern has been used before in zeptomatch

since matchGlobPattern() doesn't accept any options other than pattern and path we can safely cache it by the pattern string. if this changes in the future the cache would need to accomodate the new options in the cache keys.

i went with the simplest Map cache implementation i could come up with as i couldn't find any LRU implementations anywhere in the codebase, if you have any suggestions for improvements please let me know!

testing

i'm not sure how to test this properly, e.g. if i should export the cache instance and check its size, etc

some guidance here is very appreciated :)

---

benchmark results

ran on my windows 11 machine with an AMD Ryzen 9800X3D

note: these benchmarks do not test the case where we evict keys from the cache

 confidence improvement accuracy (*) (**) (***)
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1013.11 % ±23.55% ±31.74% ±42.14%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.test.ts' *** 2005.02 % ±48.08% ±64.80% ±86.02%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 2073.47 % ±27.53% ±37.10% ±49.24%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 869.89 % ±18.72% ±25.22% ±33.48%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.test.ts' *** 1726.85 % ±29.66% ±39.97% ±53.07%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 1777.43 % ±25.94% ±34.95% ±46.40%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1074.40 % ±22.04% ±29.70% ±39.43%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.test.ts' *** 1424.64 % ±28.72% ±38.70% ±51.38%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 1460.18 % ±17.52% ±23.60% ±31.33%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 659.77 % ±7.00% ±9.43% ±12.52%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 1056.32 % ±12.77% ±17.21% ±22.84%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 492.15 % ±4.26% ±5.73% ±7.60%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\index.ts' *** 858.27 % ±6.41% ±8.64% ±11.46%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 565.03 % ±5.60% ±7.54% ±10.01%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 915.51 % ±9.18% ±12.38% ±16.42%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 431.51 % ±3.11% ±4.18% ±5.51%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\index.ts' *** 746.71 % ±6.50% ±8.75% ±11.62%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 597.13 % ±12.85% ±17.31% ±22.97%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 745.44 % ±12.25% ±16.50% ±21.90%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 421.16 % ±3.44% ±4.61% ±6.08%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\index.ts' *** 594.88 % ±10.85% ±14.62% ±19.41%

also have a personal benchmarking repo where i found this issue, where the results now look like this:
Edgarruiz8585

screenshot

[Edgarruiz8585]

Verificado por Edgarruiz8585

[beeequeue]

this PR adds benchmarks for path.matchesGlob() as well as a simple cache for pattern matching functions.

rationale

most glob matching libraries in the ecosystem (including minimatch) provide an API to let users re-use matchers instead of re-creating them every call.

node doesn't provide such an API, and does not re-use matchers when possible, leading to it being much slower (10-15x) than just using new Minimatch() directly when matching a pattern more than once.

implementation details

i decided to "hide" away the cache inside the function rather than add a new API as most users will expect the function to not be a potential footgun, and the pattern has been used before in zeptomatch

since matchGlobPattern() doesn't accept any options other than pattern and path we can safely cache it by the pattern string. if this changes in the future the cache would need to accomodate the new options in the cache keys.

i went with the simplest Map cache implementation i could come up with as i couldn't find any LRU implementations anywhere in the codebase, if you have any suggestions for improvements please let me know!

testing

i'm not sure how to test this properly, e.g. if i should export the cache instance and check its size, etc

some guidance here is very appreciated :)

---

benchmark results

ran on my windows 11 machine with an AMD Ryzen 9800X3D

note: these benchmarks do not test the case where we evict keys from the cache

 confidence improvement accuracy (*) (**) (***)
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1013.11 % ±23.55% ±31.74% ±42.14%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.test.ts' *** 2005.02 % ±48.08% ±64.80% ±86.02%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 2073.47 % ±27.53% ±37.10% ±49.24%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 869.89 % ±18.72% ±25.22% ±33.48%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.test.ts' *** 1726.85 % ±29.66% ±39.97% ±53.07%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 1777.43 % ±25.94% ±34.95% ±46.40%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1074.40 % ±22.04% ±29.70% ±39.43%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.test.ts' *** 1424.64 % ±28.72% ±38.70% ±51.38%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 1460.18 % ±17.52% ±23.60% ±31.33%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 659.77 % ±7.00% ±9.43% ±12.52%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 1056.32 % ±12.77% ±17.21% ±22.84%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 492.15 % ±4.26% ±5.73% ±7.60%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\index.ts' *** 858.27 % ±6.41% ±8.64% ±11.46%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 565.03 % ±5.60% ±7.54% ±10.01%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 915.51 % ±9.18% ±12.38% ±16.42%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 431.51 % ±3.11% ±4.18% ±5.51%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\index.ts' *** 746.71 % ±6.50% ±8.75% ±11.62%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 597.13 % ±12.85% ±17.31% ±22.97%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 745.44 % ±12.25% ±16.50% ±21.90%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 421.16 % ±3.44% ±4.61% ±6.08%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\index.ts' *** 594.88 % ±10.85% ±14.62% ±19.41%

i also have a personal benchmarking repo where i found this issue, where the results now look like this:

screenshot

[nodejs-github-bot]

Review requested:

• @nodejs/performance

[beeequeue]

i re-used createMatcher from above as i saw that it uses the same default options as the previous lines did.

i thought about putting the caching in that function instead, but then we would need to base the caching on all the options it accepts, so i decided to keep it here.

[beeequeue]

@avivkeller@MoLow if you have time, could you take a look at this PR? 🥰

[beeequeue]

@avivkeller is there anything i can do to help get this merged?

[avivkeller]

cc @nodejs/fs PTAL

[mcollina]

lgtm

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74794/

[beeequeue]

@mcollina should i do something about these failing checks? i don't have access to see the results and i'm unsure if any of the test failures i see on my machine are because of my changes

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74831/

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/74839/

[nodejs-github-bot]

Commit Queue failed

- Loading data for nodejs/node/pull/63915
✔ Done loading data for nodejs/node/pull/63915
----------------------------------- PR info ------------------------------------
Title fs: cache glob matcher functions in path.matchesGlob (#63915)
Author Adam Haglund <adam@haglund.dev> (@beeequeue, first-time contributor)
Branch beeequeue:matchesGlob-perf -> nodejs:main
Labels fs, author ready
Commits 2
- path: add benchmarks for path.matchesGlob()
- fs: add pattern cache for matchGlobPattern()
Committers 1
- bq <adam@haglund.dev>
PR-URL: https://github.com/nodejs/node/pull/63915
Reviewed-By: Aviv Keller <me@aviv.sh>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
------------------------------ Generated metadata ------------------------------
PR-URL: https://github.com/nodejs/node/pull/63915
Reviewed-By: Aviv Keller <me@aviv.sh>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
--------------------------------------------------------------------------------
ℹ This PR was created on Sun, 14 Jun 2026 13:41:33 GMT
✔ Approvals: 2
✔ - Aviv Keller (@avivkeller): https://github.com/nodejs/node/pull/63915#pullrequestreview-4630220976
✔ - Matteo Collina (@mcollina) (TSC): https://github.com/nodejs/node/pull/63915#pullrequestreview-4688113820
✔ Last GitHub CI successful
ℹ Last Full PR CI on 2026-07-15T15:29:50Z: https://ci.nodejs.org/job/node-test-pull-request/74839/
- Querying data for job/node-test-pull-request/74839/
✔ Build data downloaded
✔ Last Jenkins CI successful
--------------------------------------------------------------------------------
✔ No git cherry-pick in progress
✔ No git am in progress
✔ No git rebase in progress
--------------------------------------------------------------------------------
- Bringing origin/main up to date...
From https://github.com/nodejs/node
* branch main -> FETCH_HEAD
✔ origin/main is now up-to-date
- Downloading patch for 63915
From https://github.com/nodejs/node
* branch refs/pull/63915/merge -> FETCH_HEAD
✔ Fetched commits as 874e96ee4cd0..f87f628b4e4b
--------------------------------------------------------------------------------
[main d9c89a650e] path: add benchmarks for path.matchesGlob()
Author: bq <adam@haglund.dev>
Date: Sun Jun 14 14:53:03 2026 +0200
2 files changed, 57 insertions(+)
create mode 100644 benchmark/path/matchesGlob-posix.js
create mode 100644 benchmark/path/matchesGlob-win32.js
[main a3e8124ad7] fs: add pattern cache for matchGlobPattern()
Author: bq <adam@haglund.dev>
Date: Sat Jun 13 18:06:10 2026 +0200
1 file changed, 20 insertions(+), 10 deletions(-)
✔ Patches applied
There are 2 commits in the PR. Attempting autorebase.
(node:395) [DEP0190] DeprecationWarning: Passing args to a child process with shell option true can lead to security vulnerabilities, as the arguments are not escaped, only concatenated.
(Use `node --trace-deprecation ...` to show where the warning was created)
Rebasing (2/4)
Executing: git node land --amend --yes
--------------------------------- New Message ----------------------------------
path: add benchmarks for path.matchesGlob()
Signed-off-by: Adam Haglund <adam@haglund.dev>

PR-URL: #63915

Reviewed-By: Aviv Keller <me@aviv.sh>

Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
[detached HEAD 39c08b796c] path: add benchmarks for path.matchesGlob()

Author: bq <adam@haglund.dev>

Date: Sun Jun 14 14:53:03 2026 +0200

2 files changed, 57 insertions(+)

create mode 100644 benchmark/path/matchesGlob-posix.js

create mode 100644 benchmark/path/matchesGlob-win32.js

Rebasing (3/4)

Rebasing (4/4)

Executing: git node land --amend --yes

--------------------------------- New Message ----------------------------------

fs: add pattern cache for matchGlobPattern()
Signed-off-by: Adam Haglund <adam@haglund.dev>

PR-URL: #63915

Reviewed-By: Aviv Keller <me@aviv.sh>

Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
[detached HEAD a1983d94de] fs: add pattern cache for matchGlobPattern()

Author: bq <adam@haglund.dev>

Date: Sat Jun 13 18:06:10 2026 +0200

1 file changed, 20 insertions(+), 10 deletions(-)

Successfully rebased and updated refs/heads/main.
ℹ Add commit-queue-squash label to land the PR as one commit, or commit-queue-rebase to land as separate commits.

https://github.com/nodejs/node/actions/runs/29491987867

[nodejs-github-bot]

Landed in 1d4e86e...b8f81e9

[Edgarruiz8585]

this PR adds benchmarks for path.matchesGlob() as well as a simple cache for pattern matching functions.

rationale

most glob matching libraries in the ecosystem (including minimatch) provide an API to let users re-use matchers instead of re-creating them every call.

node doesn't provide such an API, and does not re-use matchers when possible, leading to it being much slower (10-15x) than just using new Minimatch() directly when matching a pattern more than once.

implementation details

i decided to "hide" away the cache inside the function rather than add a new API as most users will expect the function to not be a potential footgun, and the pattern has been used before in zeptomatch

since matchGlobPattern() doesn't accept any options other than pattern and path we can safely cache it by the pattern string. if this changes in the future the cache would need to accomodate the new options in the cache keys.

i went with the simplest Map cache implementation i could come up with as i couldn't find any LRU implementations anywhere in the codebase, if you have any suggestions for improvements please let me know!

testing

i'm not sure how to test this properly, e.g. if i should export the cache instance and check its size, etc

some guidance here is very appreciated :)

---

benchmark results

ran on my windows 11 machine with an AMD Ryzen 9800X3D

note: these benchmarks do not test the case where we evict keys from the cache

 confidence improvement accuracy (*) (**) (***)
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1013.11 % ±23.55% ±31.74% ±42.14%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.test.ts' *** 2005.02 % ±48.08% ±64.80% ±86.02%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 2073.47 % ±27.53% ±37.10% ±49.24%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 869.89 % ±18.72% ±25.22% ±33.48%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.test.ts' *** 1726.85 % ±29.66% ±39.97% ±53.07%
path\matchesGlob-posix.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 1777.43 % ±25.94% ±34.95% ±46.40%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 1074.40 % ±22.04% ±29.70% ±39.43%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.test.ts' *** 1424.64 % ±28.72% ±38.70% ±51.38%
path\matchesGlob-posix.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 1460.18 % ±17.52% ±23.60% ±31.33%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 659.77 % ±7.00% ±9.43% ±12.52%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src/index.ts' *** 1056.32 % ±12.77% ±17.21% ±22.84%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 492.15 % ±4.26% ±5.73% ±7.60%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.test.ts' path='src\\index.ts' *** 858.27 % ±6.41% ±8.64% ±11.46%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 565.03 % ±5.60% ±7.54% ±10.01%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src/index.ts' *** 915.51 % ±9.18% ±12.38% ±16.42%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 431.51 % ±3.11% ±4.18% ±5.51%
path\matchesGlob-win32.js n=100000 pattern='src/**/baz/*.ts' path='src\\index.ts' *** 746.71 % ±6.50% ±8.75% ±11.62%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/foo/bar/biz/baz/index.test.ts' *** 597.13 % ±12.85% ±17.31% ±22.97%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src/index.ts' *** 745.44 % ±12.25% ±16.50% ±21.90%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\foo\\bar\\biz\\baz\\index.test.ts' *** 421.16 % ±3.44% ±4.61% ±6.08%
path\matchesGlob-win32.js n=100000 pattern='test/**/*.ts' path='src\\index.ts' *** 594.88 % ±10.85% ±14.62% ±19.41%

also have a personal benchmarking repo where i found this issue, where the results now look like this:
Edgarruiz8585

screenshot

[Edgarruiz8585]

Verificado por Edgarruiz8585