GeezerVison

Review requested:

• @nodejs/loaders

https://github.com/nodejs/node/blob/main/doc/api/module.md needs updating too, the various lines like this:

  * `format` {string|null|undefined} The format optionally supplied by the
    `resolve` hook chain

To remove null from the list.

IIUC, before this change, defaultResolve would set format to null when it can't find a better value (and load would still accept undefined nonetheless in case that's what a user hook returned). With this change, it will now be undefined (and load would still accept null nonetheless in case that's what a user hook returned). How does that make tests more precise? It seems to me it's just switching a value for another, that seems rather pointless.

I accidentally found this PR and looked a bit into the details (might still miss something). But iiuc the goal of the module detection is to allow .js files to be either CJS or ESM as opposed to current behaviour without the --experimental-detect-module that is .js are assumed to be CJS if not otherwise specified with --experimental-default-type command line option. Is this correct?

I can have a look and open a PR if I find a solution. I agree that null or undefined have equivalent explanatory value.

cc: @aduh95, @GeoffreyBooth

I accidentally found this PR and looked a bit into the details (might still miss something). But iiuc the goal of the module detection is to allow .js files to be either CJS or ESM as opposed to current behaviour without the --experimental-detect-module that is .js are assumed to be CJS if not otherwise specified with --experimental-default-type command line option. Is this correct?

#53016 has some more information, but essentially, in the end, we want it to default to CJS, but it defaults to null / undefined

CI: https://ci.nodejs.org/job/node-test-pull-request/59696/

I was actually planning on closing this favor of another PR (I need to find it then I'll link it)