[clang] [clang][Sema] Don't mark VarDecls of bindings in tuple-like decompositions as implicit (PR #100355)

[Michael Buch via cfe-commits]

Michael137 wrote:

> Thanks for the patch!
> 
> I would actually prefer option 2 here.
> 
> `isImplicit` has other uses - namely for AST matchers and refactoring tools that also rely on knowing whether something was spelled in source.
> 
> Note that I cannot comment on what make sense for debugs info, but I don't think modifying properties of the AST is reasonable, and your suggested option 2 is fine.
> 
> The structure of the AST is not stable but whoever change is would have to fix the code anyway as it's all the same project so it's a non-issue (as long as you add sufficient test coverage)

Thanks all for the input! Happy to go with option 2 instead. I agree that changing the AST here just for `DW_AT_artificial` emission does feel a bit drastic/is not obviously the right thing to do.

https://github.com/llvm/llvm-project/pull/100355