Peter Jaško

At the very least, Rollup correctly tree-shakes it. See [repl](https://rollupjs.org/repl/?version=3.22.0&shareable=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). I agree it is unergonomic to have to manually `import * as R from 'remeda'` since typescript won't automatically guess...

@JoviDeCroock I suppose that if Preact's signals are to be widely-used, some conventions should emerge. Such as, if a library exposes a component that takes signals as props, whose responsibility...

And it also gets me thinking, I haven't experimented as much with `useSignalEffect()` so let's put that aside for a while and assume signals work perfectly until you use `useComputed()`....

Made some changes after feedback from software wallet devs

Heads-up: There are conflicts with main so I'm about to rebase

Heads-up: Rebasing again to fix conflict

Certain values do avoid the error, but unfortunately I need `define: {global: 'globalThis'}`, i.e. replacing an identifier with another identifier, not a string literal.

Has this been fixed?

It seems like the library already has types defined in `docs/*.rst`, how are those being generated? Could type stubs be generated the same way?

Is topological sorting out of scope? Seems like the most useful feature of `pnpm run -r` and `yarn workspaces foreach`