Annotate internal function IDs

[r0qs]

fixes #13425

[nikola-matic]

This is a fairly large test case, which in turn generates an even larger AST JSON. Could this be split into multiple test cases? (I'm fine with also leaving this one as an all encompassing one as well, but additional smaller test cases would be nice)

[r0qs]

This test was extracted from:

solidity/test/libsolidity/analysis/FunctionCallGraph.cpp

Line 1199 in 0aa8515

BOOST_AUTO_TEST_CASE(indirect_calls)

But I'm not sure how to split it, I mean, the test is only to check it the internalFunctionID is exported for internal functions, but I'm not really sure of all the cases that this will be trigger, do you have any suggestions of corner cases?

I could also just simplify it and have only (f)();

[nikola-matic]

I'm not even talking about edge cases per se - e.g. you could have a separate case with just a free function (like free1 in the current test case), which would yield a much smaller AST. Same thing goes for a minimal contract with one internal function, and one public/external, where the AST will again be quite small, and the internal function will have the internalFunctionID member, whereas the public/external won't. And so on, and so on.

[ekpyron]

Actually, for the current AST tests we have, any new tests comes with like 1000 lines of boilerplate, so splitting this up into several cases actually makes things even worse :-). In the future, we should really have targetted AST tests that just selectively check for a subset of fields - me being too lazy to write something like that is actually the blocker on nice stuff like #13378...

So if you're up for it, you can just do that for this PR - but I'm also fine with keeping the large test case here and dealing with that separately (it should be a separate PR no matter what anyways). Maybe better to move this forward like it is and then build a nicer way to test json ASTs only afterwards to be sure to finally get this out in the next release (Truffle has been asking for these function IDs for ages now).

[ekpyron]

Apart from the comments above, this is mostly ready I think and mainly missing a changelog entry.

[ekpyron]

We can also try and draw @haltman-at's attention to the binaries in this PR, resp. to the AST in https://github.com/ethereum/solidity/blob/96dee72fe5b5fb6e68fbe25cdf00c3470a2ce794/test/libsolidity/ASTJSON/ast_internal_function_id_export.json (and the internalFunctionID fields in that) to confirm that this is a good enough fix for #13425 for the needs of the Truffle debugger.

[ekpyron]

I think we can squash this and then merge this (although I'd wait for a bit to see if @haltman-at is around for confirming that this in fact solves #13425 for them). Also nice that pregenerating the IDs like this actually ends up being much simpler than what we used to do on-the-fly.

[r0qs]

@haltman-at we would love to hear your comments on this before merging. If the PR indeed fixes the issue for your use case.

[ekpyron]

I think this should solve the issue, so I'm merging this.

[haltman-at]

(FWIW, I was on sabbatical for a few weeks and so didn't have a chance to test this out, but I'll assume you all know what you're doing. :) Will still try to test it before release still if I have a chance I guess...)