[Tests] Amend typos and extraneous <Set /> generic typings in router tests

[Philzen]

Just a handful of test name and typescript amendments – discovered while skimming over https://github.com/redwoodjs/redwood/pull/10464/files

[Philzen]

@Tobbe i'm afraid you're not going to like this (packages/router/src/tests/router.test.tsx):

#11739 was still fine – it's also what i would love to see in the next v8 patch release as it fixes what broke during the migration from 7 → 8.

Starting with #11756 the typing all went downhill 🙈
Issue there

• It was in theory a nice idea of me to use keyof AvailableRoutes for autocompletion, but poorly executed as of course it only worked with my locally generated types in the .redwood folder. So any route name occurences would either have to be @ts-ignored ignored now in the main project or we'd need to find another way to generate those route typings.
• Then this change (i believe) then broke the nested types

Not sure if this is easily fixable or if we maybe should revert starting from #11756 for the time being and try another day. Just wanted to quickly share this as a heads up.

[Tobbe]

Arrrg! Router typings! What a mess 🤬 Why is this to difficult to get right 😰

I'm down to revert if we think that's the best way forward

[Tobbe]

Trying to figure out what PRs to revert

#11756
#11767
#11769

@Philzen Does that look like the complete list to you?

[Philzen]

@Philzen Does that look like the complete list to you?

I just double-checked – everything after #11739 (which is OK) adds 8 red squiggles to packages/router/src/tests/router.test.tsx. So for revert that would be

#11769
#11768
#11767
#11756

… in that order of revert should work.

Why is this to difficult to get right 😰

Something about AvailableRoutes apparently is more complicated than it looks – at least that is the PR where things went south.

I'm down to revert if we think that's the best way forward

Well, at least we have a document of our joint failure in the git history now 🤣 – we can always start over from there.
Going forward, i feel it is vital to include TS typecheck into the CI to avoid this kind of disappointing hindsight surprises. I realize it's impossible to get the whole repo to pass out-of-the-box, so we should start with the ones that are safe, fixing one by one and including them in the TS check job as we go along.

Then we would at least be sure typings can only get better and never get worse / broken again 😎

[Tobbe]

PRs have been reverted

#11902 reverts #11769
#11903 reverts #11768
#11904 reverts #11767
#11905 reverts #11756

[Tobbe]

I actually think this was better the way it was.
Had this been an it-test instead I would have agreed with your change. But with test I think "test: Give correct..." reads better

(But I really wish all tests used it())

[Tobbe]

With all the type change PRs reverted I think I want to keep this as it was. What's your thought on that, @Philzen?

[Philzen]

At the end of the day, these are completely redundant – there is no more or less TS errors in these files with or without them.

But let me double check if it changes anything regarding auto-completion – although even then, they would be just noise in the test, as they don't serve any purpose concerning stuff that would be able to come up in a CI.

My thought is to find a use case where they are actually required and implement a test for that, whilst removing those generics where they aren't.