AaronBallman wrote:

> In that case, we could perhaps try a modular rollout—provided the maintainers 
> of the respective modules don't mind the calls becoming more cumbersome.
> 
> However, this does present a practical issue: if more than half of the 
> sub-project maintainers disagree, the use of separate maintenance method 
> would lead to code bloat—in which case, it would actually be better not to 
> remove the API in the first place.

Yeah, that's why I think we need more discussion about this, there's not a 
clearly correct answer yet.

I don't think we should make a sweeping change to remove calls to 
`getQualifiedNameAsString()` due to the churn. (It's still fine to make 
targeted changes which remove it though, of course!) I also don't think we 
should remove the FIXME yet either. I think this probably deserves broader 
community discussion (and research into current performance characteristics of 
the call) because this is a case where Clang needs may differ from tooling 
needs, which is a bit more of a negotiation and perhaps that means the Clang 
Area Team should host a meeting for interested parties to discuss it. So we 
might as well get the discussion started on Discourse first before jumping into 
calls about it. However, given that nothing is on fire in this space, that 
might be more effort than folks are interested in taking on, so it's also fine 
to say "let's just leave the situation alone for now."

https://github.com/llvm/llvm-project/pull/218264
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to