3.1.58 breaks usage in Node.js worker

[toyobayashi]

minimum repro:

https://github.com/toyobayashi/emscripten-node-worker-3158-repro
https://github.com/toyobayashi/emscripten-node-worker-3158-repro/actions/runs/8827891977

issue 1: Cannot find correct .wasm path when run emscripten generated JavaScript in a Node.js Worker.
issue 2: The empty *.worker.js is written in cwd, not the same directory where .js exists.
issue 3: Pthread worker will read incorrect .wasm file again even if I specified locateFile when -sSTRICT=1

[sbc100]

Is this perhaps a duplicate of #21826?

The core misunderstand there seems to be that pthreads need to be able to locate the wasm file. In fact pthreads recieve the wasm module via postMessage so they should never be trying to locate or read the wasm file themselves.

I believe the issue that caused some confusion in #21826 was that locateFile is still being called on worker startup, even though the worker is never using the result of that call. I think I will update the code so workers never actaully call locateFile to avoid this confusion.

[toyobayashi]

not sure whether issue 3 is duplicated, but I confirm 1 and 2 are not related to #21826.

I mean run emscripten js in my own Node.js worker (the worker is emscripten's main thread), please take a look the repro code and CI test, it does break something.

[toyobayashi]

Issue 1 has nothing to do with locateFile, after I experienced issue 1, I tried to use locateFile to solve it then experienced issue 3.

[sbc100]

So we have a fix for (2) in #21830 and (3) in #21829.

That just leaves (1) which still needs some investigation.