Improve error reporting: Missing = on type declaration

[tboby]

This is my attempt to implement #1121, leaning on the start made in #6603.

This is my first contribution, so feedback appreciated!

I'm not quite sure what's happening with the xlf files. Is it possible a previous commit on master forgot to generate them?

[smoothdeveloper]

@tboby

I'm not quite sure what's happening with the xlf files. Is it possible a previous commit on master forgot to generate them?

It is better to consider they are in .gitignore (despite they aren't), and not check them in unless explicitly asked during the review.

[abelbraaksma]

Not 100% sure, but maybe the xlf files had something to do with: #9641

I do know that they need to be rebuild when texts change, so that the translation team gets notified to translate the new texts. That that means that there's a bunch of changes at once seems to me to be expected.

[tboby]

I've:

• rebased onto main
• removed the xlf files as I wasn't sure if I was supposed to leave them in
• updated the error message and parser naming as suggested

[tboby]

I've amended this to include the type name in the error message.
However, this required me to take the head or tail of the list of idents, which was odd, as the spec implies a type def's type name should be an ident:
type-name : attributesopt accessopt ident typar-defnsopt
not a long-ident (which includes dots).

The tyconDefn: parser definition I edit here eventually just uses path which is SyntaxTree.LongIdentWithDots. If you try and use dots in a type name you do seem to get an error, but this is caught by the TypeChecker which emits "Invalid type extension".

I may be missing something, but unless someone lets me know better I will write this up into an issue and consider having a go at changing this. I assume it is better to catch a spec syntax error in the parser than in the type checker?

[cartermp]

I assume it is better to catch a spec syntax error in the parser than in the type checker?

Yes, absolutely :)

@dsyme any thoughts on the spec-y question?

[cartermp]

Thanks for this work, it is great!