add docstring for new kwargs in code_typed

[Roger-luo]

I find the docstring for interp and world is missing in 1.6+, so I just add them here.

---

I also add an example to help people remember whether the signature is Tuple{Float64, Float64} or (Float64, Float64) I find I mix them up quite often

[KristofferC]

Are these two (Base.get_world_counter() and Core.Compiler.NativeInterpreter) referenced from anywhere else? Otherwise, it seems a bit odd to have it in a docstring like this.

[Roger-luo]

I don't think they are documented in other places yet, should I just remove them from here then?

[Roger-luo]

Or should I also add a docstring for them?

[vchuravy]

Please add a docstring for them as well

[Roger-luo]

I removed the references to internal functions, hope this looks good now.

[KristofferC]

I still claim that these new keyword arguments (world, interp) are not documented anywhere else, are considered internal, and should therefore not be a part of the official documentation. If anything, they should be in something like devdocs since they require knowledge and usage of Julia internals.

[simeonschaub]

We do document Base.invoke_in_world though, so it seems reasonable to me to document the world argument here and probably Base.get_world_counter() as well.

[KristofferC]

We do document Base.invoke_in_world though,

Documented means "exist in the manual". A docstring on an internal function is pretty much just a glorified comment (this is one reason why I don't think docstrings on internal functions are ideal because people constantly get confused about if it is an internal function or not). The manual (excluding the devdocs) should be consistent, you should be able to figure out what you can pass for every argument for every function just by reading it. And this does not mean that invoke_in_world should be added to the manual and made public.

[Roger-luo]

OK what should I do here? maybe move it to dev doc?

[Roger-luo]

I think the main issue here is that I find not mentioning the new keyword arguments (world, interp) in docstring makes things confusing when using it. what about if we just mention these two keyword argument should be considered for who knows internal? I'm planning to submit other similar docstring updates to improve the experience programming compiler plugins, so would be nice to have some clarification on this type of documentation.

[vchuravy]

I would add in the docstring a separate heading denotation these Kearns as internal and referencing the devdocs

[Roger-luo]

@vchuravy like this?

what is "Kearns"? you mean Kwargs?

[DilumAluthge]

@staticfloat Some strange Buildkite failures here.

[vchuravy]

Is this not just because this branch needs to be rebased?

[DilumAluthge]

Ah, maybe that's it. @Roger-luo can you rebase on the latest master?

[Roger-luo]

any idea why linux64 fails after merging master into this branch?