Remove need for use_attr and use_proc

[gui1117]

implements first part of #3

what it does:
import_tokens macros call themself instead of calling their related inner_sig
forward_tokens_inner_internal will add a keyword at the start of the token stream.
if import_tokens see the keyword at the start of the token stream then it call their inner_sig which is declared inlined.

tests are successful

TODO:

• update docs
• discuss the keyword and whereas we should expose a forward_token which doesn't add this extra token in order not to break usage of forward_token
• is it correct to assume usage of extra is equivalent to having been called by an attribute macro, or should we pass along this information in another way.

[sam0x17]

This is amazing, didn't think this would be possible without a stack of macros calling each other. I will have to go over carefully to understand how you got around that!

[sam0x17]

for context, extra right now is a sort of dumping ground for things that need to be passed from the initial macro call to the final macro call that were added on as later features from the initial version. It's sort of horrible how I pass it and use string manipulation to escape it, really it should be done on the export_tokens macro itself but this was a convenient way to quickly add a few features.

[sam0x17]

I am also OK with changing / breaking the original signature of forward_tokens if needed -- I don't think anyone is using it directly and I already see this as a 0.4x thing so OK to have a breaking change

[sam0x17]

So yeah if you can get the cargo fmt to pass I can merge this and we can tackle the next pieces

[gui1117]

@sam0x17 what I wrote here looks a bit abusive to me, but as long as we use the extra token stuff only for attribute macros we are good.
but maybe it should be refactored together with extra token feature.

otherwise we could have forward_tokens_to_attribute_macro and forward_tokens_to_proc_macro.
the implementation would redirect to forward_tokens_inner(attr: true, ...) and forward_tokens_inner(attr: false, ...). and then same for forward_tokens_internal.

[sam0x17]

Yeah also if I remember correctly the main blocker to having a more general thing like a tt here is anything but an ident or a literal limits what types of position you can use forward_tokens stuff with because of how decl macros work. This is the same reason I have to do that crazy manual interpolation of idents with :: instead of just a path in the export tokens macro. Extra as a string works in all positions.

Maybe a tuple like this would be better though:

("extra item 1", "extra item 2", "extra item 3", "extra item 4")

I don't think that would mess up the position capabilities at all, and at least then we wouldn't have to do string interpolation and escaping

[sam0x17]

But yes I think it is fine to expand to a proc macro here, and I'd be happy to merge it this way

[gui1117]

yes formatted, I only have the little concern above #5 (comment) but the PR is ready.

I will attend buidl asia, I won't work on the other points soon I think

[sam0x17]

OK I will see if I can get the other parts working and update the docs 👍🏻