include generic bodies in allowMetaTypes

[metagn]

fixes #19848

Not sure why this wasn't the case already. The if cl.allowMetaTypes: return line below for tyFromExpr was added 10 years ago. Hopefully it was just negligence?

[Araq]

Why do we need allowMetaTypes at all?

[metagn]

I'm not too familiar with it, in general it seems to be to produce tyGenericInst for unresolved param types instead of tyGenericInvocation/tyGenericBody for easier type checking?

For this instance it's triggered here in the only call to prepareMetatypeForSigmatch. The code around it is strongly implying it's a hack and should be moved to liftParamType but liftParamType also uses it to convert tyGenericBody to a tyGenericInst in a tyCompositeTypeClass and tyGenericInvocation to a resolved concept type.

I think @zerbina has managed to remove it in case they have any insight, as far as I can tell it would mostly involve adding more logic to typeRel for tyGenericBody and tyGenericInvocation to mirror what tyGenericInst does but maybe this comes with some compromises/technical challenges.

[zerbina]

I'm not too familiar with it, in general it seems to be to produce tyGenericInst for unresolved param types instead of tyGenericInvocation/tyGenericBody for easier type checking?

That, and, like you said, also for the tyCompositeTypeClass lifting.

I think @zerbina has managed to remove it in case they have any insight

It's been a while since I've worked on that area of the compiler, and I don't remember all the details. There's an issue I created back then that goes into more detail and also references the various prerequisites for removing allowMetaTypes. The relevant PRs should contain more information about the various challenges encountered on the way.

The existence of allowMetaTypes has far reaching consequences, and its removal made significant stability improvements possible.

[github-actions]

Thanks for your hard work on this PR!
The lines below are statistics of the Nim compiler built from 1befb8d

Hint: mm: orc; opt: speed; options: -d:release
173486 lines; 8.075s; 654.527MiB peakmem