fix: derived reactivity and perf regressions

[hmnd]

fixes sveltejs/kit#15059, fixes #17342, fixes #17111, fixes #17323, fixes #17352

#17105 resulted in a few regressions to how deriveds with no dependencies are treated. At worst, it broke reactivity for deriveds in branch effects, but I realized it also casued perf regressions like #17342. Both are now addressed, with tests to match.

To summarize, the issues I've fixed are:

• deriveds with no deps were always getting set as MAYBE_DIRTY when they should have been CLEAN, resulting in excessive re-evaluation
• deriveds couldn't reconnect when read in a branch effect (because effect_tracking() is false there)
• child deriveds weren't being reconnected when read by a connected parent derived (5.43.4+ $derived doesn't update #17352)

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
• Prefix your PR title with feat:, fix:, chore:, or docs:.
• This message body should clearly illustrate what problems it solves.
• Ideally, include a test that fails without this PR but passes with it.
• If this PR changes code within packages/svelte/src, add a changeset (npx changeset).

Tests and linting

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

[changeset-bot]

🦋 Changeset detected

Latest commit: 5ef1d01

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

This PR includes changesets to release 1 package

Name	Type
svelte	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

[svelte-docs-bot]

Preview: https://svelte-dev-git-preview-svelte-17362-svelte.vercel.app/

[github-actions]

Playground

pnpm add https://pkg.pr.new/svelte@17362

[hmnd]

@paoloricciuti would love if you could review/merge this one too :)

As you can see, this regression is the source of many head aches!

[paoloricciuti]

@hmnd yeah I already took a look...I think the fix makes sense but I'm not super familiar with the logic of the connected deriveds as I would like so I would not feel confident to review and merge this on my own. I've already shared it in the maintainers chat yesterday tho so hopefully we can get this merged soon.

Thanks for your effort 🧡

[Rich-Harris]

Made some progress — swapped out the unit test of the bug in #17352 for a component-based test, since these tend to be a little easier to reason about and debug in our sandbox.

Would love to see if we can whittle it down further, but will come back to it later

[Rich-Harris]

alright, think we're good to go now. thank you!

[hmnd]

Thank you for your work refining this to a usable fix! Glad I could contribute something meaningful, and hope to be able to be of more help in the future.