vgvassilev wrote:

> > With .PolishForDeclaration=true, there are NO final specifiers (which is 
> > what we want to produce forward decls in IWYU)
> 
> This is actually a regression in this PR, and it breaks the clangd test added 
> here: 
> [9f57b65](https://github.com/llvm/llvm-project/commit/9f57b65a272817752aa00e2fb94154e6eed1d0ec)
>  (the patch that originally led to double `final`s).
> 
> EDIT: It looks like `prettyPrintAttributes` is only intended for semantic 
> attributes, whereas clangd wanted to preserve as-written `final` keyword. So 
> I wonder if this change needs to be tweaked a little: 
> https://github.com/llvm/llvm-project/pull/88600/files#diff-81d69bc555945d6582a758e0c094ff870cbc38697501ad7415694ee30c567dbfL1085

I've added a fix for your example. Can you provide a test case for the clangd 
use-case?

https://github.com/llvm/llvm-project/pull/88600
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to