-
Notifications
You must be signed in to change notification settings - Fork 30k
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(esm): leverage loaders when resolving subsequent loaders #43772
feat(esm): leverage loaders when resolving subsequent loaders #43772
Conversation
9044b9c
to
b234023
Compare
lib/internal/process/esm_loader.js
Outdated
for (let i = 0; i < customLoaders.length; i++) { | ||
const customLoader = customLoaders[i]; | ||
|
||
// Importation must be handled by internal loader to avoid poluting userland | ||
const keyedExportsSublist = await internalEsmLoader.import( | ||
[customLoader], | ||
pathToFileURL(cwd).href, | ||
ObjectCreate(null), | ||
); | ||
|
||
await internalEsmLoader.addCustomLoaders(keyedExportsSublist); | ||
ArrayPrototypePushApply(keyedExportsList, keyedExportsSublist); | ||
} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
A) How is implementing “loaders applying to subsequent loaders” so simple, and B) does this break any existing tests?
Looking at this code, there’s no way I would’ve guessed that this was using existing loaders as part of resolving and loading subsequent loaders. Could you add a comment (or several) explaining that that’s what’s happening here, and how?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
My naive thinking is that since loader resolution goes through a loader pipeline (even if it doesn't initially contain any loader but the default one), all we need is to progressively add the loaders to it so that they are used by subsequent resolutions.
A few tests don't seem to pass yet, I'm searching why.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'm a bit confused by the approach here, so I reviewed only the first part of this PR. I would expect something like this to suffice:
- Import ambient loaders (via internal ESMLoader)
- Add ambient loaders to internal ESMLoader
- Import lay loaders (via internal ESMLoader, thus affected by ambient loaders added above)
- Add ambient loaders to public ESMLoader
- Add lay loaders to public ESMLoader
main...JakobJingleheimer:node:feat/add-esm-ambient-loaders
Could you help me understand why the extra complexity is necessary?
EDIT: If the above is sufficient, I'd be happy to pick this up (seems pretty quick to knock out), but also don't want to step on your toes.
lib/internal/process/esm_loader.js
Outdated
const keyedExportsSublist = await internalEsmLoader.import( | ||
[customLoader], | ||
pathToFileURL(cwd).href, | ||
ObjectCreate(null), | ||
); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Why loop over customLoaders
just to feed each one as a wrapped list to ESMLoader::import()
(which already supports multiple specifiers)?
I would expect simply:
const keyedExportsSublist = internalESMLoader(customLoaders);
internalESMLoader.addCustomLoaders(keyedExportsSublist);
ArrayPrototypePushApply(loaders, keyedExportsSublist);
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This diff makes it so that we load loaders one at a time (because each one must be used to resolve/loader the subsequent ones).
Why the wrapper: without it, import
makes an early return which shortcuts the pairing.
Your implementation seems to split the loaders in two, but doesn't entirely answer the requirement to "leverage loaders when resolving subsequent loaders". For instance, with your diff, only the following would work:
And the following wouldn't work:
In other words, |
Ahhh, yes: ambient loaders need to be imported and also added to the list of hooks before the next is imported (so the previous hook runs during the subsequent's import) 🤔 |
hello :) |
The plan is to land this once it’s ready and after #44710. |
From the last loaders meeting notes:
|
@arcanis Can you please rebase this on |
70f2987
to
7f30aae
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'm almost in disbelief how simple this is 🙌 Seeing that, I don't anticipate any clashes with off-threading.
I think the ESM doc just needs to be updated and this is good to go.
EDIT: And perhaps some code docs about the new interactive behaviour between loaders?
@@ -306,7 +306,7 @@ class ESMLoader { | |||
|
|||
/** | |||
* Collect custom/user-defined hook(s). After all hooks have been collected, | |||
* calls global preload hook(s). | |||
* the global preload hook(s) must be called. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Maybe a bit of explanation as to why this can't happen automatically? (since it previously did)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Cf this thread:
Because otherwise, since
addCustomLoaders
is now called multiple times (once for each loader), it would lead topreload
being called multiple times, which in turn would execute the global preload hook multiple times, which would crash (you can try reverting this change; a test crashes).
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Sure, something like that'll do 🙂 (I'm suggesting adding a code comment)
+1 to backport request; this functionality is necessary for use of Yarn PNP alongside https://github.com/DataDog/import-in-the-middle as per yarnpkg/berry#4044 |
@nodejs/loaders We should probably prepare a single backport PR with all recent ESM-loader changes. |
Still planning on backporting to 18.x @GeoffreyBooth / @arcanis ? |
Splitting the backport question into an Issue here #48042 |
This was missed for backport into 18.17.0 somehow. |
PR-URL: #43772 Reviewed-By: James M Snell <jasnell@gmail.com> Reviewed-By: Jacob Smith <jacob@frende.me> Reviewed-By: Geoffrey Booth <webadmin@geoffreybooth.com>
Thanks for backporting y'all. |
Notable changes: deps: * (SEMVER-MINOR) update uvwasi to 0.0.19 (Node.js GitHub Bot) #49908 * (SEMVER-MINOR) upgrade npm to 10.0.0 (npm team) #49423 doc: * add new TSC members (Michael Dawson) #48841 * move and rename loaders section (Geoffrey Booth) #49261 esm: * use import attributes instead of import assertions (Antoine du Hamel) #50140 * --experimental-default-type flag to flip module defaults (Geoffrey Booth) #49869 * unflag import.meta.resolve (Guy Bedford) #49028 * move hook execution to separate thread (Jacob Smith) #44710 * leverage loaders when resolving subsequent loaders (Maël Nison) #43772 lib: * (SEMVER-MINOR) add api to detect whether source-maps are enabled (翠 / green) #46391 * (SEMVER-MINOR) add tracing channel to diagnostics_channel (Stephen Belanger) #44943 src: * (SEMVER-MINOR) add cjs_module_lexer_version base64_version (Jithil P Ponnan) #45629 stream: * use bitmap in readable state (Benjamin Gruenbaum) #49745 test_runner: * (SEMVER-MINOR) accept `testOnly` in `run` (Moshe Atlow) #49753 * (SEMVER-MINOR) add junit reporter (Moshe Atlow) #49614 * (SEMVER-MINOR) expose location of tests (Colin Ihrig) #48975 * (SEMVER-MINOR) add shards support (Raz Luvaton) #48639 * (SEMVER-MINOR) add initial draft for fakeTimers (Erick Wendel) #47775 test_runner, cli: * (SEMVER-MINOR) add --test-concurrency flag (Colin Ihrig) #49996 tls: * (SEMVER-MINOR) add ALPNCallback server option for dynamic ALPN negotiation (Tim Perry) #45190 vm: * (SEMVER-MINOR) use import attributes instead of import assertions (Antoine du Hamel) #50141 wasi: * (SEMVER-MINOR) updates required for latest uvwasi version (Michael Dawson) #49908 PR-URL: TODO
Notable changes: deps: * (SEMVER-MINOR) update uvwasi to 0.0.19 (Node.js GitHub Bot) #49908 * (SEMVER-MINOR) upgrade npm to 10.2.3 (npm team) #50531 doc: * move and rename loaders section (Geoffrey Booth) #49261 esm: * use import attributes instead of import assertions (Antoine du Hamel) #50140 * --experimental-default-type flag to flip module defaults (Geoffrey Booth) #49869 * unflag import.meta.resolve (Guy Bedford) #49028 * move hook execution to separate thread (Jacob Smith) #44710 * leverage loaders when resolving subsequent loaders (Maël Nison) #43772 lib: * (SEMVER-MINOR) add api to detect whether source-maps are enabled (翠 / green) #46391 * (SEMVER-MINOR) add tracing channel to diagnostics_channel (Stephen Belanger) #44943 src: * (SEMVER-MINOR) add cjs_module_lexer_version base64_version (Jithil P Ponnan) #45629 stream: * use bitmap in readable state (Benjamin Gruenbaum) #49745 test_runner: * (SEMVER-MINOR) accept `testOnly` in `run` (Moshe Atlow) #49753 * (SEMVER-MINOR) add junit reporter (Moshe Atlow) #49614 * (SEMVER-MINOR) expose location of tests (Colin Ihrig) #48975 * (SEMVER-MINOR) add shards support (Raz Luvaton) #48639 * (SEMVER-MINOR) add initial draft for fakeTimers (Erick Wendel) #47775 test_runner, cli: * (SEMVER-MINOR) add --test-concurrency flag (Colin Ihrig) #49996 tls: * (SEMVER-MINOR) add ALPNCallback server option for dynamic ALPN negotiation (Tim Perry) #45190 vm: * (SEMVER-MINOR) use import attributes instead of import assertions (Antoine du Hamel) #50141 PR-URL: #50953
Notable changes: deps: * (SEMVER-MINOR) update uvwasi to 0.0.19 (Node.js GitHub Bot) #49908 * (SEMVER-MINOR) upgrade npm to 10.2.3 (npm team) #50531 doc: * move and rename loaders section (Geoffrey Booth) #49261 esm: * use import attributes instead of import assertions (Antoine du Hamel) #50140 * --experimental-default-type flag to flip module defaults (Geoffrey Booth) #49869 * unflag import.meta.resolve (Guy Bedford) #49028 * move hook execution to separate thread (Jacob Smith) #44710 * leverage loaders when resolving subsequent loaders (Maël Nison) #43772 lib: * (SEMVER-MINOR) add api to detect whether source-maps are enabled (翠 / green) #46391 * (SEMVER-MINOR) add tracing channel to diagnostics_channel (Stephen Belanger) #44943 src: * (SEMVER-MINOR) add cjs_module_lexer_version base64_version (Jithil P Ponnan) #45629 stream: * use bitmap in readable state (Benjamin Gruenbaum) #49745 test_runner: * (SEMVER-MINOR) accept `testOnly` in `run` (Moshe Atlow) #49753 * (SEMVER-MINOR) add junit reporter (Moshe Atlow) #49614 * (SEMVER-MINOR) expose location of tests (Colin Ihrig) #48975 * (SEMVER-MINOR) add shards support (Raz Luvaton) #48639 * (SEMVER-MINOR) add initial draft for fakeTimers (Erick Wendel) #47775 test_runner, cli: * (SEMVER-MINOR) add --test-concurrency flag (Colin Ihrig) #49996 tls: * (SEMVER-MINOR) add ALPNCallback server option for dynamic ALPN negotiation (Tim Perry) #45190 vm: * (SEMVER-MINOR) use import attributes instead of import assertions (Antoine du Hamel) #50141 PR-URL: #50953
PR-URL: nodejs/node#43772 Reviewed-By: James M Snell <jasnell@gmail.com> Reviewed-By: Jacob Smith <jacob@frende.me> Reviewed-By: Geoffrey Booth <webadmin@geoffreybooth.com>
Notable changes: deps: * (SEMVER-MINOR) update uvwasi to 0.0.19 (Node.js GitHub Bot) nodejs/node#49908 * (SEMVER-MINOR) upgrade npm to 10.2.3 (npm team) nodejs/node#50531 doc: * move and rename loaders section (Geoffrey Booth) nodejs/node#49261 esm: * use import attributes instead of import assertions (Antoine du Hamel) nodejs/node#50140 * --experimental-default-type flag to flip module defaults (Geoffrey Booth) nodejs/node#49869 * unflag import.meta.resolve (Guy Bedford) nodejs/node#49028 * move hook execution to separate thread (Jacob Smith) nodejs/node#44710 * leverage loaders when resolving subsequent loaders (Maël Nison) nodejs/node#43772 lib: * (SEMVER-MINOR) add api to detect whether source-maps are enabled (翠 / green) nodejs/node#46391 * (SEMVER-MINOR) add tracing channel to diagnostics_channel (Stephen Belanger) nodejs/node#44943 src: * (SEMVER-MINOR) add cjs_module_lexer_version base64_version (Jithil P Ponnan) nodejs/node#45629 stream: * use bitmap in readable state (Benjamin Gruenbaum) nodejs/node#49745 test_runner: * (SEMVER-MINOR) accept `testOnly` in `run` (Moshe Atlow) nodejs/node#49753 * (SEMVER-MINOR) add junit reporter (Moshe Atlow) nodejs/node#49614 * (SEMVER-MINOR) expose location of tests (Colin Ihrig) nodejs/node#48975 * (SEMVER-MINOR) add shards support (Raz Luvaton) nodejs/node#48639 * (SEMVER-MINOR) add initial draft for fakeTimers (Erick Wendel) nodejs/node#47775 test_runner, cli: * (SEMVER-MINOR) add --test-concurrency flag (Colin Ihrig) nodejs/node#49996 tls: * (SEMVER-MINOR) add ALPNCallback server option for dynamic ALPN negotiation (Tim Perry) nodejs/node#45190 vm: * (SEMVER-MINOR) use import attributes instead of import assertions (Antoine du Hamel) nodejs/node#50141 PR-URL: nodejs/node#50953
PR-URL: nodejs/node#43772 Reviewed-By: James M Snell <jasnell@gmail.com> Reviewed-By: Jacob Smith <jacob@frende.me> Reviewed-By: Geoffrey Booth <webadmin@geoffreybooth.com>
Notable changes: deps: * (SEMVER-MINOR) update uvwasi to 0.0.19 (Node.js GitHub Bot) nodejs/node#49908 * (SEMVER-MINOR) upgrade npm to 10.2.3 (npm team) nodejs/node#50531 doc: * move and rename loaders section (Geoffrey Booth) nodejs/node#49261 esm: * use import attributes instead of import assertions (Antoine du Hamel) nodejs/node#50140 * --experimental-default-type flag to flip module defaults (Geoffrey Booth) nodejs/node#49869 * unflag import.meta.resolve (Guy Bedford) nodejs/node#49028 * move hook execution to separate thread (Jacob Smith) nodejs/node#44710 * leverage loaders when resolving subsequent loaders (Maël Nison) nodejs/node#43772 lib: * (SEMVER-MINOR) add api to detect whether source-maps are enabled (翠 / green) nodejs/node#46391 * (SEMVER-MINOR) add tracing channel to diagnostics_channel (Stephen Belanger) nodejs/node#44943 src: * (SEMVER-MINOR) add cjs_module_lexer_version base64_version (Jithil P Ponnan) nodejs/node#45629 stream: * use bitmap in readable state (Benjamin Gruenbaum) nodejs/node#49745 test_runner: * (SEMVER-MINOR) accept `testOnly` in `run` (Moshe Atlow) nodejs/node#49753 * (SEMVER-MINOR) add junit reporter (Moshe Atlow) nodejs/node#49614 * (SEMVER-MINOR) expose location of tests (Colin Ihrig) nodejs/node#48975 * (SEMVER-MINOR) add shards support (Raz Luvaton) nodejs/node#48639 * (SEMVER-MINOR) add initial draft for fakeTimers (Erick Wendel) nodejs/node#47775 test_runner, cli: * (SEMVER-MINOR) add --test-concurrency flag (Colin Ihrig) nodejs/node#49996 tls: * (SEMVER-MINOR) add ALPNCallback server option for dynamic ALPN negotiation (Tim Perry) nodejs/node#45190 vm: * (SEMVER-MINOR) use import attributes instead of import assertions (Antoine du Hamel) nodejs/node#50141 PR-URL: nodejs/node#50953
“Notable change” summary by @GeoffreyBooth:
Original introduction by @arcanis:
This PR is a follow-up to the Ambient Loaders proposal. Since in the last meetings we weren’t too sure whether “ambient loaders” shouldn’t just be the default loaders behaviour, I implemented it this way (but can change it later if needed).
I left a few assets to quickly try the system; without this PR, the second
--loader
would fail becausexxx/*
wouldn’t be resolvable:cc @nodejs/loaders @cspotcode