`resolveId` is called before `buildStart` with enforce: 'pre' about vitest HOT 5 OPEN

Hi! rollup-plugin-node-externals author here.

I was not aware of this and the linked discussion until @chandu filed Septh/rollup-plugin-node-externals#30 which I fixed by not relying on buildStart to initialize my variables. Simple enough.

However I do believe that this is a bug that should be fixed in Vitest because the order in which the hooks are called is clearly defined by Rollup's documentation and diverging from that order is wrong (and my plugin might not be the only one affected).

from vitest.

It looks like buildStart is still called, but the issue is that resolveId is called before buildStart when you have enforce: "pre".
I also made a repro here:
https://github.com/hi-ogawa/reproductions/tree/main/vitest-5616-plugin-build-start

> vitest

@resolveId(pre) { source: 'vitest' }
@resolveId(pre) { source: 'vitest' }
@buildStart(pre)
@buildStart(normal)

 DEV  v1.5.1 /home/projects/wmaivylmj.github
...

I remembered someone mentioned that buildStart is not working, but it looks like the same issue where resolveId being getting called first:

• #5479

from vitest.

This resolveId seems to be the trigger, which is called during configureServer and before buildStart

Quick logging from plugin shows something like this:

@configureServer(pre)
@resolveId(pre) { source: 'vitest' }
@resolveId(pre) { source: 'vitest' }
@configureServer(normal)
@buildStart(pre)
@buildStart(normal)

@sagargurtu Can you explain what is your use case? The linked discussion mentioned rollup-plugin-node-externals. I think this is technically what Vitest shouldn't do, but I'm not sure if there is a quick way around.
Depending on the use case, there might be a workaround on plugin side.

from vitest.

@hi-ogawa While I could potentially move my initialization code to config() or the function that returns the plugin object, I agree with @Septh as many other existing plugins could break because of this issue.

from vitest.