feat: Add ssr.noExternal = true option

[jplhomer]

Description

This PR adds a ssr.noExternal = true option which disables all externals during the SSR build process. This is effectively the same as manually adding all dependencies of a package to ssr.noExternal.

This is helpful when building for Workers environments like Cloudflare Workers, where require is not supported at all.

This PR also adds a check to the resolve plugin to ensure the user doesn't attempt to import any Node.js built-ins, as those wouldn't be supported in the Workers environment.

Additional context

This functionality was discussed in a comment thread here: #4230 (comment)

Another option I considered was to imply this behavior from ssr.target = 'webworker'. However, that would be a breaking change, and I think it's better to be explicit in this case (maybe somebody does build a webworker runtime someday that supports require - who knows?)

Tests: I'd love to know the most effective way to test this behavior. Since it's exclusive to Worker environments, and it's only related to builds, it's tough to use the playground tests. I'd also want to test the output of the file, meaning unit tests aren't really the right way to go (unless someone can point me to an example).

UPDATE: Tests have been added. See #4490 (comment)

I also changed the name from bundle to bundleAll.

UPDATE 2: Changed to ssr.noExternal = true per feedback.

---

What is the purpose of this pull request?

• Bug fix
• New Feature
• Documentation update
• Other

Before submitting the PR, please make sure you do the following

• Read the Contributing Guidelines.
• Read the Pull Request Guidelines and follow the Commit Convention.
• Check that there isn't already a PR that solves the problem the same way to avoid creating a duplicate.
• Provide a description in this PR that addresses what the PR is solving, or reference the issue that it solves (e.g. fixes #123).
• Ideally, include relevant tests that fail without this PR but pass with it.

[benmccann]

Maybe we should call this bundleAll?

I had been thinking that it could be really nice to rename ssr.noExternal to ssr.bundle at some point. If @brillout's plan to make everything external at some point comes to fruition we wouldn't need the ssr.external option anymore - just an option to bundle certain libraries. But even if we never do this rename, I think that bundleAll is more descriptive of what the option does since Vite is already bundling but just isn't bundling everything

[brillout]

@benmccann You're right naming it bundleAll is probably better.

[frandiox]

I think that bundleAll is more descriptive of what the option does since Vite is already bundling but just isn't bundling everything

That was my original reasoning of having bundle-all vs bundle. But I guess both work well as long as it's documented.

[jplhomer]

bundleAll it is! 🎉

I've also added a new playground test, ssr-webworker. It boots a custom worker server using miniflare. Only a couple assertions have been added, but it could be a useful place to debug worker-related issues in the future.

[brillout]

Neat 👌. From a high-level perspective, the code LGTM.

[patak-dev]

Looks good to me 👍🏼
We'll discuss this new feature with the team next week.

[Shinigami92]

Today we discussed with the team that we want to prevent more and more options and would like to reuse noExternal instead of bundleAll.
So we could do something like noExternal: true or noExternal: '*'.

[patak-dev]

Let's go with noExternal: true, as it seems to properly imply that there isn't anything externalized. Just a detail, but I think that this good because these APIs may change in the future (example of a proposal to do so). Since bundling everything is part of this API, it seems better to me to avoid new options that we may need to deprecate in a month.

[jplhomer]

Good call on the option usage. I've updated this PR to use ssr.noExternal = true.

[Shinigami92]

Could you revert your yarn.lock? I think it has nothing todo with the PR.

[Shinigami92]

Still changes in yarn.lock 🤔 Maybe you need a rebase?