perf(resolve): refactor package.json handling for deep imports #12461

bluwy · 2023-03-17T09:21:19Z

Description

Partial revert of #5665. That PR was not exports friendly and we had to fix it in #10371, but it mixes both "root" package.json and "nearest" package.json, which needs to be handled separately.

This PR refactors in a way that pkg passed to resolveDeepImport and resolvePackageEntry is always the "root" package.json. If further down the chain a resolve logic needs the "nearest" package.json, they can resolve it themselves.

This change shouldn't break resolvePackageEntry (since that's root import only), but changes resolveDeepImport slightly, where:

If "root" package.json has no exports field, the browser field (if available) will be derive from the "root" package.json (now), instead of the "nearest" package.json (before)

I think the above is a bug fix though.

More details about `browser`

According to https://github.com/defunctzombie/package-browser-field-spec#alternate-main---basic:

All paths for browser fields are relative to the package.json file location (and usually project root as a result).

It implies that package.json other than "root" should be checked too, but if we respect it, that means we have to check all package.json that's part of the current file path, which is a lot of work even if the browser field was never used.

I think it's fine for now to check the browser field based on "root" package.json only as that's the more common case.

Additional context

No benchmarks 😬 But I noticed this optimization as we're consistently trying to resolve package.json for long deep import paths due to:

vite/packages/vite/src/node/plugins/resolve.ts

Lines 717 to 725 in 535c8c5

    
           const nearestPkgId = [...possiblePkgIds].reverse().find((pkgId) => { 
        
             nearestPkg = resolvePackageData( 
        
               pkgId, 
        
               basedir, 
        
               preserveSymlinks, 
        
               packageCache, 
        
             )! 
        
             return nearestPkg 
        
           })!

What is the purpose of this pull request?

Bug fix
New Feature
Documentation update
Other

Before submitting the PR, please make sure you do the following

Read the Contributing Guidelines.
Read the Pull Request Guidelines and follow the PR Title Convention.
Check that there isn't already a PR that solves the problem the same way to avoid creating a duplicate.
Provide a description in this PR that addresses what the PR is solving, or reference the issue that it solves (e.g. fixes #123).
Ideally, include relevant tests that fail without this PR but pass with it.

stackblitz · 2023-03-17T09:21:23Z

Run & review this pull request in StackBlitz Codeflow.

patak-dev · 2023-03-17T09:49:14Z

/ecosystem-ci run

vite-ecosystem-ci · 2023-03-17T09:49:39Z

📝 Ran ecosystem CI: Open

suite	result
astro	✅ success
histoire	❌ failure
iles	✅ success
ladle	✅ success
laravel	✅ success
marko	✅ success
nuxt	❌ failure
previewjs	✅ success
qwik	❌ failure
rakkas	✅ success
sveltekit	✅ success
vite-plugin-ssr	✅ success
vite-plugin-react	✅ success
vite-plugin-react-pages	✅ success
vite-plugin-react-swc	✅ success
vite-plugin-svelte	✅ success
vite-plugin-vue	✅ success
vite-setup-catalogue	✅ success
vitepress	❌ failure
vitest	✅ success
windicss	✅ success

patak-dev · 2023-03-17T10:18:04Z

Qwik, Nuxt, and Histoire are also currently failing on main (same errors).

But Vitepress against main was successful, so there may be a regression for it 👀

bluwy · 2023-03-17T12:49:38Z

I ran the VitePress test locally, and I can get the same error, but it's not always reproducible. When they do fail, I can confirm that the VitePress page has an empty <div id="app"></div> which would fail the test. But I'm not sure how that happens.

I can confirm when running some repros that this PR improves the resolve speed slightly, so it's not slowing down the tests that causes it to timeout. Not really sure what to do here though, we could maybe test again when we start the minor.

I'm guessing it's some sort of race condition, since running that failing test directly via pnpm test-init is less likely to cause the fail.

patak-dev · 2023-03-17T12:55:00Z

cc @brc-dd in case you have some time to help us debugging this in VitePress before 4.3

brc-dd · 2023-03-17T13:12:58Z

Yeah, I'll look into this, it's probably just a flaky test IG 👀 Even in our repo, we sometimes have to rerun jobs.

brc-dd · 2023-03-17T13:21:31Z

@patak-dev Hey, can you try rerunning the CI for vitepress and see if it's consistently failing for a specific test? If it's random or some race condition, then feel free to go ahead with the PR, I don't think vite has anything to do with that.

Ah, ok nvm. You guys already triggered it 3 times.

brc-dd · 2023-03-17T14:27:51Z

Seems like some issue with one of our deps, appears to be working after bumping them: https://github.com/brc-dd/vite-ecosystem-ci/actions/runs/4448482529/attempts/1, https://github.com/brc-dd/vite-ecosystem-ci/actions/runs/4448586970/jobs/7811590602

patak-dev · 2023-03-17T14:38:12Z

/ecosystem-ci run vitepress

vite-ecosystem-ci · 2023-03-17T14:38:42Z

📝 Ran ecosystem CI: Open

suite	result
vitepress	❌ failure

brc-dd · 2023-03-17T14:43:45Z

Failed 🫠

patak-dev · 2023-03-17T15:07:31Z

/ecosystem-ci run vitepress

vite-ecosystem-ci · 2023-03-17T15:08:02Z

📝 Ran ecosystem CI: Open

suite	result
vitepress	❌ failure

bluwy · 2023-03-17T15:30:11Z

Yeah it looks like something in this PR is causing it to happen more frequently, but I can't tell where. Thanks for looking into this though @brc-dd!

patak-dev · 2023-03-17T15:55:05Z

/ecosystem-ci run vitepress

vite-ecosystem-ci · 2023-03-17T15:55:31Z

📝 Ran ecosystem CI: Open

suite	result
vitepress	✅ success

patak-dev · 2023-03-17T18:25:05Z

/ecosystem-ci run vitepress

vite-ecosystem-ci · 2023-03-17T18:25:32Z

📝 Ran ecosystem CI: Open

suite	result
vitepress	✅ success

perf(resolve): refactor package.json handling for deep imports

ce344f3

bluwy added p4-important Violate documented behavior or significantly improves performance (priority) performance Performance related enhancement labels Mar 17, 2023

patak-dev added this to the 4.3 milestone Mar 17, 2023

patak-dev approved these changes Mar 21, 2023

View reviewed changes

patak-dev merged commit 596b661 into main Mar 21, 2023

patak-dev deleted the perf-resolve-pkg-deep branch March 21, 2023 06:09

btea mentioned this pull request Apr 26, 2023

fix(resolve): deep import resolvedId error #13010

Merged

9 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

perf(resolve): refactor package.json handling for deep imports #12461

perf(resolve): refactor package.json handling for deep imports #12461

bluwy commented Mar 17, 2023 •

edited

Loading

stackblitz bot commented Mar 17, 2023

patak-dev commented Mar 17, 2023

vite-ecosystem-ci bot commented Mar 17, 2023 •

edited

Loading

patak-dev commented Mar 17, 2023

bluwy commented Mar 17, 2023 •

edited

Loading

patak-dev commented Mar 17, 2023

brc-dd commented Mar 17, 2023

brc-dd commented Mar 17, 2023 •

edited

Loading

brc-dd commented Mar 17, 2023 •

edited

Loading

patak-dev commented Mar 17, 2023

vite-ecosystem-ci bot commented Mar 17, 2023 •

edited

Loading

brc-dd commented Mar 17, 2023

patak-dev commented Mar 17, 2023

vite-ecosystem-ci bot commented Mar 17, 2023 •

edited

Loading

bluwy commented Mar 17, 2023

patak-dev commented Mar 17, 2023

vite-ecosystem-ci bot commented Mar 17, 2023 •

edited

Loading

patak-dev commented Mar 17, 2023

vite-ecosystem-ci bot commented Mar 17, 2023 •

edited

Loading

	const nearestPkgId = [...possiblePkgIds].reverse().find((pkgId) => {
	nearestPkg = resolvePackageData(
	pkgId,
	basedir,
	preserveSymlinks,
	packageCache,
	)!
	return nearestPkg
	})!

perf(resolve): refactor package.json handling for deep imports #12461

perf(resolve): refactor package.json handling for deep imports #12461

Conversation

bluwy commented Mar 17, 2023 • edited Loading

Description

Additional context

What is the purpose of this pull request?

Before submitting the PR, please make sure you do the following

stackblitz bot commented Mar 17, 2023

patak-dev commented Mar 17, 2023

vite-ecosystem-ci bot commented Mar 17, 2023 • edited Loading

patak-dev commented Mar 17, 2023

bluwy commented Mar 17, 2023 • edited Loading

patak-dev commented Mar 17, 2023

brc-dd commented Mar 17, 2023

brc-dd commented Mar 17, 2023 • edited Loading

brc-dd commented Mar 17, 2023 • edited Loading

patak-dev commented Mar 17, 2023

vite-ecosystem-ci bot commented Mar 17, 2023 • edited Loading

brc-dd commented Mar 17, 2023

patak-dev commented Mar 17, 2023

vite-ecosystem-ci bot commented Mar 17, 2023 • edited Loading

bluwy commented Mar 17, 2023

patak-dev commented Mar 17, 2023

vite-ecosystem-ci bot commented Mar 17, 2023 • edited Loading

patak-dev commented Mar 17, 2023

vite-ecosystem-ci bot commented Mar 17, 2023 • edited Loading

bluwy commented Mar 17, 2023 •

edited

Loading

vite-ecosystem-ci bot commented Mar 17, 2023 •

edited

Loading

bluwy commented Mar 17, 2023 •

edited

Loading

brc-dd commented Mar 17, 2023 •

edited

Loading

brc-dd commented Mar 17, 2023 •

edited

Loading

vite-ecosystem-ci bot commented Mar 17, 2023 •

edited

Loading

vite-ecosystem-ci bot commented Mar 17, 2023 •

edited

Loading

vite-ecosystem-ci bot commented Mar 17, 2023 •

edited

Loading

vite-ecosystem-ci bot commented Mar 17, 2023 •

edited

Loading