Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

feat: prerender & analyse in worker rather than subprocess to support Deno #9919

Merged
merged 3 commits into from
May 16, 2023

Conversation

fernandolguevara
Copy link
Contributor

@fernandolguevara fernandolguevara commented May 13, 2023

use workers instead of child_process

related:
denoland/deno#17248

@benmccann would you be able to review this PR?

@changeset-bot
Copy link

changeset-bot bot commented May 13, 2023

🦋 Changeset detected

Latest commit: 333e7ed

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

This PR includes changesets to release 1 package
Name Type
@sveltejs/kit Patch

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

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

Copy link
Member

@Rich-Harris Rich-Harris left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

this LGTM! thank you. just had one small observation, inline.

I'm not sure I totally understand how this affects Deno though — it looks like child_process and worker_threads are both implemented. Is there some nuance I'm missing?

packages/kit/src/utils/fork.js Outdated Show resolved Hide resolved
@benmccann benmccann changed the title feat(fork): use workers feat: prerender in worker rather than subprocess to support Deno May 13, 2023
@benmccann
Copy link
Member

I'm not sure I totally understand how this affects Deno though — it looks like child_process and worker_threads are both implemented. Is there some nuance I'm missing?

It sounds to me like Deno implements child_process, but not communication between child and parent process and for that recommend using workers - or at least that's what I got from this article. @bartlomieju would you be able to confirm if that's correct?

Co-authored-by: Rich Harris <hello@rich-harris.dev>
@fernandolguevara fernandolguevara changed the title feat: prerender in worker rather than subprocess to support Deno feat: prerender & analyse in worker rather than subprocess to support Deno May 13, 2023
@bartlomieju
Copy link

Thanks for the ping @benmccann.

It sounds to me like Deno implements child_process, but not communication between child and parent process

That's true, we currently don't have support for the child_process.send() API. Unfortunately our polyfill for the worker_threads module is broken at the moment, but given that you have a need for it, we can prioritize fixing it and get it done next week. Does that sounds reasonable to you?

@benmccann
Copy link
Member

Thanks! Yeah, we can wait to merge this on our side then. I'm also fine going either the worker_threads or child_process.send() route if one is easier than the other

@bartlomieju
Copy link

Workers would definitely be easier - we just need to update once-working code. I'll prioritize this work over the next week and get back to you once it's released (I'll target to release in Deno v1.33.4 next week).

@dominikg
Copy link
Member

iirc we have been using child_process to ensure resources are properly cleaned up (eg db connections). Is that still the case with worker?

@bartlomieju
Copy link

iirc we have been using child_process to ensure resources are properly cleaned up (eg db connections). Is that still the case with worker?

@dominikg I believe so - workers use separate OS threads, so once the worked is closed/terminated the resources will be released back to the OS.

@Rich-Harris
Copy link
Member

This code guarantees that everything is correctly shut down — if it wasn't, the tests would never complete:

// this code is here to make sure that we kill the process
setInterval(() => {
console.log('process is still alive');
}, 5000);

So given that the tests are passing I'm confident that it works correctly.

@Rich-Harris Rich-Harris merged commit 2e6da94 into sveltejs:master May 16, 2023
@github-actions github-actions bot mentioned this pull request May 16, 2023
leonardoadame pushed a commit to leonardoadame/Affiliate-tech that referenced this pull request May 17, 2023
* feat: Add a speedier script tag for prerendered redirects (sveltejs#9911)

* feat: Add a script redirect to prerendered pages

* changeset

* Update .changeset/hungry-rocks-hunt.md

Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>

* fix test

* feat: Update string escaping

* Update .changeset/hungry-rocks-hunt.md

Co-authored-by: Conduitry <git@chor.date>

---------

Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
Co-authored-by: Conduitry <git@chor.date>

* fix: avoid inlining raw/url CSS imports (sveltejs#9925)

* fix: use transformRequest for CSS modules

* just avoid raw or url

* test and changeset

* use parsed query

* remove only

* Update packages/kit/test/apps/basics/test/cross-platform/test.js

* drive by test speed up

* feat: prerender & analyse in worker rather than subprocess to support Deno (sveltejs#9919)

* feat(fork): use workers

* Create hot-actors-hope.md

* Update packages/kit/src/utils/fork.js

Co-authored-by: Rich Harris <hello@rich-harris.dev>

---------

Co-authored-by: Ben McCann <322311+benmccann@users.noreply.github.com>
Co-authored-by: Rich Harris <hello@rich-harris.dev>

* avoid using isMainThread, since it interacts poorly with vitest (sveltejs#9941)

Co-authored-by: Rich Harris <git@rich-harris.dev>

* chore: bump vite and devalue (sveltejs#9933)

* bump vite and devalue

* update templates

* merge master

* fix test

---------

Co-authored-by: Rich Harris <git@rich-harris.dev>

* chore: uvu -> vitest for create-svelte tests (sveltejs#9910)

* chore: uvu -> vitest for create-svelte tests

* format

* concurrency (sveltejs#9921)

* realised we werent typechecking this file, found some errors. fixed

* Wait for beforeAll hook to complete

* simplify

* exclude create-svelte/template files from prettier, so that we can emit correctly formatted templates

* remove unused file

* Revert "exclude create-svelte/template files from prettier, so that we can emit correctly formatted templates"

This reverts commit aa188d4.

---------

Co-authored-by: Ben McCann <322311+benmccann@users.noreply.github.com>
Co-authored-by: Rich Harris <git@rich-harris.dev>

* feat: unshadow `form` and `data` in `enhance` (sveltejs#9902)

* feat: Un-shadow `data` and `form` in `enhance`, warn about future deprecation in dev

* changeset

* snek

* Update .changeset/odd-crews-own.md

Co-authored-by: Ben McCann <322311+benmccann@users.noreply.github.com>

* Update packages/kit/test/apps/dev-only/package.json

Co-authored-by: Ben McCann <322311+benmccann@users.noreply.github.com>

* am not smart

* still not smart

* oops

* oof

* add deprecation notice

---------

Co-authored-by: Ben McCann <322311+benmccann@users.noreply.github.com>
Co-authored-by: Rich Harris <git@rich-harris.dev>

* fix: Set loader: { '.wasm': 'copy' } in esbuild config in `adapter-cloudflare-workers` (sveltejs#9940)

* fix: Set loader: { '.wasm': 'copy' } in esbuild config in `adapter-cloudflare-workers`

Copies WASM files in Cloudflare instead of trying to load them.

Related to sveltejs#9909

* Create brave-peaches-buy.md

* Update packages/adapter-cloudflare-workers/index.js

* format

---------

Co-authored-by: Rich Harris <git@rich-harris.dev>
Co-authored-by: Rich Harris <hello@rich-harris.dev>

* fix: Flaky test (sveltejs#9947)

* fix: Flaky test

* reuse locator

* add semi

* drive by test speed up

* another classic

* oh my god brain, y u so bad

---------

Co-authored-by: gtmnayan <50981692+gtm-nayan@users.noreply.github.com>
Co-authored-by: gtmnayan <gtmnayan@gmail.com>

* remove envVarsInUse (sveltejs#9942)

Co-authored-by: Rich Harris <git@rich-harris.dev>

* fix: type `vitePlugin` in config (sveltejs#9946)

* fix: type `vitePlugin` in config

* changeset

* fix: Set `loader: { '.wasm': 'copy' }` in esbuild config in `adapter-vercel` (sveltejs#9944)

* fix: Enable wasm copy in adapter-vercel

* Create olive-rings-eat.md

---------

Co-authored-by: Rich Harris <richard.a.harris@gmail.com>

* feat: crawl URLs in `<meta>` tags (sveltejs#9900)

* Crawl social-image urls during prerender

* Formatting & Linting

* Format changeset & added exhaustive list of crawlable urls

* Changed severity to minor as described in sveltejs#5228

* Added support for `property` attribute & limited valid names to just social tags

* More tests

* Better changeset message - I'm indecisive

* Update .changeset/thirty-garlics-tan.md

Co-authored-by: Ben McCann <322311+benmccann@users.noreply.github.com>

* simplify

* simplify

* Removed redundant data-sanitation

* DRY out

---------

Co-authored-by: Ben McCann <322311+benmccann@users.noreply.github.com>
Co-authored-by: Rich Harris <git@rich-harris.dev>

* Version Packages (sveltejs#9893)

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>

* feat: support AWS via SST in adapter-auto (sveltejs#9874)

* [feat] support AWS via SST in adapter-auto

* Sync

* Delete 95-adapter-aws-sst.md

* Update .changeset/rotten-ducks-tan.md

---------

Co-authored-by: Rich Harris <hello@rich-harris.dev>

* Version Packages (sveltejs#9953)

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>

* only cache response if response has cache-control header (sveltejs#9885)

* only cache response if response has cache-control header

* add changeset

* Version Packages (sveltejs#9955)

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>

* fix: ensure styles are loaded in dev mode for routes containing special characters (sveltejs#9894)

* Fix loading styles for routes containing special characters in dev mode.

SvelteKit doesn't decode special characters in pathnames when loading CSS modules in dev mode, resulting in an error:

Internal server error: Failed to load url /src/routes/(special)/hinnap%C3%A4ring/+page.svelte?svelte=&type=style&lang.css=&inline= (resolved id: /src/routes/(special)/hinnap%C3%A4ring/+page.svelte?svelte&type=style&lang.css). Does the file exist?

Actual path:

/src/routes/(special)/hinnapäring/

Fix by using decodeURI on the url.pathname when loading CSS modules.

* Create stale-houses-yell.md

* decodeURL inside if

* add test

---------

Co-authored-by: Ben McCann <322311+benmccann@users.noreply.github.com>
Co-authored-by: Rich Harris <git@rich-harris.dev>

* feat: Warn users when submitting forms with files but no `enctype="multipart/form-data"` (sveltejs#9888)

* fix: Package name keeps me from filtering with pnpm

* feat: Warn users when submitting a form containing a file without the correct enctype

* changeset

* Update packages/kit/src/runtime/app/forms.js

Co-authored-by: gtmnayan <50981692+gtm-nayan@users.noreply.github.com>

* style

* moar style tweaks

* better test skip

* Update .changeset/tasty-llamas-relate.md

Co-authored-by: Ben McCann <322311+benmccann@users.noreply.github.com>

* DRY out

* only warn once per submit

* Update .changeset/tasty-llamas-relate.md

Co-authored-by: Rich Harris <richard.a.harris@gmail.com>

---------

Co-authored-by: gtmnayan <50981692+gtm-nayan@users.noreply.github.com>
Co-authored-by: Ben McCann <322311+benmccann@users.noreply.github.com>
Co-authored-by: Rich Harris <git@rich-harris.dev>
Co-authored-by: Rich Harris <richard.a.harris@gmail.com>

* Version Packages (sveltejs#9957)

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>

* security: Stop automatically adding URLs from server-side `load` `fetch` calls to dependencies (sveltejs#9945)

* feat: Add `dangerZone` config

* breaking: Don't implicitly track deps in server-side fetch

* changeset

* bein dumb

* fix: Server load invalidation

* fix: Write config for server

* fix: test

* unsurprisingly i am dumb

* docs: Clarify difference between server `fetch` and universal `fetch`

* tests

* rename to trackServerFetches

---------

Co-authored-by: Rich Harris <git@rich-harris.dev>

* Version Packages (sveltejs#9963)

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>

---------

Co-authored-by: S. Elliott Johnson <sejohnson@torchcloudconsulting.com>
Co-authored-by: Simon H <5968653+dummdidumm@users.noreply.github.com>
Co-authored-by: Conduitry <git@chor.date>
Co-authored-by: gtmnayan <50981692+gtm-nayan@users.noreply.github.com>
Co-authored-by: Fernando López Guevara <fernando.lguevara@gmail.com>
Co-authored-by: Ben McCann <322311+benmccann@users.noreply.github.com>
Co-authored-by: Rich Harris <hello@rich-harris.dev>
Co-authored-by: Rich Harris <richard.a.harris@gmail.com>
Co-authored-by: Rich Harris <git@rich-harris.dev>
Co-authored-by: Conner <conner@semicognitive.com>
Co-authored-by: gtmnayan <gtmnayan@gmail.com>
Co-authored-by: Loris Sigrist <43482866+LorisSigrist@users.noreply.github.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: Frank <frank@sst.dev>
Co-authored-by: Frank Dumont <fj.dumont@gmail.com>
Co-authored-by: Reio Remma <reio@mrstuudio.ee>
@pboguslawski
Copy link

Please take a look at #9989 - seems that using workers require also env cleanup.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

6 participants