HazardyKnusperkeks wrote:
> > I would do this instead:
> > ```diff
> > --- a/clang/lib/Format/TokenAnnotator.cpp
> > +++ b/clang/lib/Format/TokenAnnotator.cpp
> > @@ -2336,7 +2336,7 @@ private:
> > for (FormatToken *Previous = Current.Previous;
> > Previous && Previous->isOneOf(tok::star, tok::amp);
> > Previous = Previous->Previous) {
> > - Previous->setType(TT_PointerOrReference);
> > + Previous->setFinalizedType(TT_PointerOrReference);
> > }
> > if (Line.MustBeDeclaration &&
> > Contexts.front().ContextType != Context::CtorInitializer) {
> > ```
> >
> >
> > IMO, this is more conservative and serves a catch-all for uncovered edge
> > cases.
>
> Also, it would obsolete #212856, which caused the regression in #222132.
I don't like the idea of code which is not used/necessary for any of our tests
to pass. And while there are still some cases not covered, I think we have a
good base for annotating those tokens.
This code was useful at some point, but I think it became obsolete and no one
registered it. In this case we have invalid/incomplete code, but there's no
real chance that these already annotated tokens would be pointers, so
overwriting that seems like a bad idea to me.
https://github.com/llvm/llvm-project/pull/220080
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
