module: execute --import sequentially

[aduh95]

Fixes: #50427

[nodejs-github-bot]

Review requested:

• @nodejs/loaders

[Uzlopak]

If one of the modules has a top level await you have to load the modules in order, but if not you could load them unordered.
So it is a fix which can have performance regressions?

[JakobJingleheimer]

So it is a fix which can have performance regressions?

I think it's more accurate to say this has performance implications.

[anonrig]

I'm -1 on this. Is it really worth preserving the order when sequential awaiting has a huge impact on performance?

[aduh95]

has a huge impact on performance?

You say huge, how big are we talking about?

[anonrig]

has a huge impact on performance?

You say huge, how big are we talking about?

That's a good question that deserves a benchmark 👍

[GeoffreyBooth]

I’m -1 on this. Is it really worth preserving the order when sequential awaiting has a huge impact on performance?

Yes, because the current implementation doesn’t achieve the desired effect. If the first --import registers hooks to handle TypeScript, the second --import of a TypeScript file should work. It doesn’t matter how much slower it becomes; the current behavior is broken.

If the user wants parallelized loading, they can put a bunch of import statements into a file and --import that.

Convinced by @GeoffreyBooth's comments

[aduh95]

has a huge impact on performance?

You say huge, how big are we talking about?

That's a good question that deserves a benchmark 👍

I don't think it's that good of a question, it was mostly rhetorical. The change has literally no impact on folks who are not using --import, and the impact of folks using lots of --import will depend a lot on what's inside the imported modules. I don't think we can come up with a benchmark showing regressions unless it's very artificial (i.e. no real-world use case would be impacted), but happy to be proven wrong.

--import is an experimental API, performance consideration is not a strong enough reason to block that PR from landing IMO (unless you're saying it will slow down users who are not using it, but I don't think it's the case).

[giltayar]

I’m -1 on this. Is it really worth preserving the order when sequential awaiting has a huge impact on performance?

@GeoffreyBooth supplied one reason it has to be sequential. Another reason is the "loader registration use case" as outlined in #50427 (and which was how this bug was found), where if the user writes --import loader-a --import loader-b then they want to register the loaders to be registered in that specific order because loader chain order is important: some loaders have to be chained in just the right order.

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/55363/

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/55366/

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/55367/

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/55377/

[nodejs-github-bot]

Commit Queue failed

- Loading data for nodejs/node/pull/50474
✔  Done loading data for nodejs/node/pull/50474
----------------------------------- PR info ------------------------------------
Title      module: execute `--import` sequentially (#50474)
   ⚠  Could not retrieve the email or name of the PR author's from user's GitHub profile!
Branch     aduh95:--import-serial -> nodejs:main
Labels     process, author ready, needs-ci, commit-queue-squash
Commits    2
 - module: execute `--import` sequentially
 - Apply suggestions from code review
Committers 2
 - Antoine du Hamel 
 - GitHub 
PR-URL: https://github.com/nodejs/node/pull/50474
Fixes: https://github.com/nodejs/node/issues/50427
Reviewed-By: Jacob Smith 
Reviewed-By: Michaël Zasso 
Reviewed-By: Moshe Atlow 
------------------------------ Generated metadata ------------------------------
PR-URL: https://github.com/nodejs/node/pull/50474
Fixes: https://github.com/nodejs/node/issues/50427
Reviewed-By: Jacob Smith 
Reviewed-By: Michaël Zasso 
Reviewed-By: Moshe Atlow 
--------------------------------------------------------------------------------
   ⚠  Commits were pushed since the last approving review:
   ⚠  - Apply suggestions from code review
   ℹ  This PR was created on Mon, 30 Oct 2023 10:21:43 GMT
   ✔  Approvals: 3
   ✔  - Jacob Smith (@JakobJingleheimer): https://github.com/nodejs/node/pull/50474#pullrequestreview-1703780454
   ✔  - Michaël Zasso (@targos) (TSC): https://github.com/nodejs/node/pull/50474#pullrequestreview-1703829531
   ✔  - Moshe Atlow (@MoLow) (TSC): https://github.com/nodejs/node/pull/50474#pullrequestreview-1705908328
   ✔  Last GitHub CI successful
   ℹ  Last Full PR CI on 2023-11-01T05:42:53Z: https://ci.nodejs.org/job/node-test-pull-request/55377/
- Querying data for job/node-test-pull-request/55377/
   ✔  Last Jenkins CI successful
--------------------------------------------------------------------------------
   ✔  Aborted `git node land` session in /home/runner/work/node/node/.ncu

https://github.com/nodejs/node/actions/runs/6718540805

[nodejs-github-bot]

Landed in a3e09b3