Possible source generator fix

[Briaoeuidhtns]

I came across this recent comment that suggests a workaround for dependent source generators. I can look into it myself, but I want to make sure it's not something that's been tried already.

[christianhelle]

@Briaoeuidhtns No, I haven't tried to re-run the generators that should be aware of the generated code, from your this generator

It's worth a try!

[christianhelle]

@Briaoeuidhtns Did it work?

[christianhelle]

@Briaoeuidhtns I got a bit curious, and I also gave it a shot. As far as I can see, this would only work if Refit exposed InterfaceStubGenerator in the Refit package. This is, unfortunately, not the case right now. We could try to duplicate it, but that would be a bit counter-productive as we will need to keep it in sync with Refits implementation

[Briaoeuidhtns]

Thanks, I hadn't had a chance to check yet. Is it worth making an issue for Refit to see if they'd expose it?

[christianhelle]

We should investigate more before we involve the Refit team. I would like to see a working proof-of-concept that was suggested in dotnet/roslyn#57239 (comment)

When I gave it a shot, I quickly ran into a wall regarding compilation units being immutable and the code that Refitter generated wasn't picked up by the InterfaceStubGenerator correctly.

If I (or someone else, hint hint...) get to a working proof-of-concept after some minor tweaks to InterfaceStubGenerator without breaking existing Refit functionality then we could do a pull request on Refit