fix: $env/*/private throws for client-side imports on Windows and Tailwind doesn't freak out

[elliott-with-the-longest-name-on-github]

Closes #5723 and closes #5747.

The paths for both Vite and Rollup's module graphs needed to be normalized, along with the illegal import paths.

Because Tailwind needs to analyze basically every file in your project, Vite thinks it imports all of them, meaning it'll almost always import $env/*/private, causing our helpful module analysis to be, well, unhelpful. In dev, we now ignore non-js|ts|svelte files. Rollup doesn't have this problem (because it doesn't have to worry about HMR), so we can be stricter and just analyze all imports. I think this strikes a happy medium. "Provide armor against footguns in dev, and provide even better armor when we try to go to prod."

Please don't delete this checklist! Before submitting the PR, please make sure you do the following:

• It's really useful if your PR references an issue where it is discussed ahead of time. In many cases, features are absent for a reason. For large changes, please create an RFC: https://github.com/sveltejs/rfcs
• This message body should clearly illustrate what problems it solves.
• Ideally, include a test that fails without this PR but passes with it.

Tests

• Run the tests with pnpm test and lint the project with pnpm lint and pnpm check

Changesets

• If your PR makes a change that should be noted in one or more packages' changelogs, generate a changeset by running pnpm changeset and following the prompts. All changesets should be patch until SvelteKit 1.0

[changeset-bot]

🦋 Changeset detected

Latest commit: 00fbfeb

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 1 package

Name	Type
@sveltejs/kit	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[dominikg]

is it possible to use vite's normalizePath here? https://vitejs.dev/guide/api-plugin.html#path-normalization

[elliott-with-the-longest-name-on-github]

I was hoping someone more knowledgeable would come along and make my day.

[benmccann]

I'm surprised none of the browser tests failed on Windows. Maybe we need one that uses $env/*/private?

[elliott-with-the-longest-name-on-github]

I'm surprised none of the browser tests failed on Windows. Maybe we need one that uses $env/*/private?

The problem is that these are buildtime/dev-time errors, meaning we'd need a test project dedicated to attempting to build and ensuring the build fails. There are a number of things that'd be useful for -- I'm happy to put an implementation together, but probably not for this PR?

[benmccann]

I'm not sure you'd need anything too special or a separate test app. If you fail the build on one of the existing test apps it should turn the CI red

[elliott-with-the-longest-name-on-github]

I'm not sure you'd need anything too special or a separate test app. If you fail the build on one of the existing test apps it should turn the CI red

Yes, but we're trying to verify that the build does fail, and the build needs to not fail in order for the rest of the tests to be able to run. Hence, the "metaproject" that tries to build purposefully-failing projects and verifies that they actually do fail.

[benmccann]

Oh, right. Yeah, that'd be kind of a pain to test

[elliott-with-the-longest-name-on-github]

Rich, there's another env var bug I want to work on tonight, but it might step on this one's toes. Easier to merge this one first, if you think it looks good.

[elliott-with-the-longest-name-on-github]

Heads up @Rich-Harris -- I went ahead and fixed the other bug in this PR as well. It was a pretty simple change.

[elliott-with-the-longest-name-on-github]

Okay, once we see the green christmas tree list we should be GTM @Rich-Harris

[Rich-Harris]

we need a snake_case lint rule!

[Rich-Harris]

actually i rescind my approval until i'm confident that the firefox tests are just flaking