krisb added inline comments.
================
Comment at: clang/lib/Parse/ParseDecl.cpp:3080
+ // For instance this marked as unavailable:
+ // class __attribute((unavailable)) UnavailableClass;`
+ auto RemoveAccessCheckingDiagnostics = [&TemplateInfo, this]() {
----------------
aorlov wrote:
> krisb wrote:
> > Basically, if `__attribute((unavailable))` should trigger the error for any
> > use of an unavailable class, we have it already broken.
> > For example, for this code clang-12 doesn't produce any diagnostics:
> > ```
> > class __attribute((unavailable)) X {
> > template <typename T> class __attribute((unavailable)) Y {};
> > };
> > class __attribute((unavailable)) A {
> > class __attribute((unavailable)) C {};
> > };
> > template <> class X::Y<A::C> {};
> > ```
> > So, I don't see much sense in inventing something new to workaround only
> > the cases that come with this patch. It's better to either fix it globally
> > or leave it broken atm with the corresponding FIXME.
> Anyway, I tried to remove the access diagnostics only. Who knows which
> diagnostics may be needed. Or you strongly prefer using
> `SuppressAccessChecks` instead?
Basically, I'd prefer consistency. Having two different ways of doing (almost)
the same things doesn't look good to me. But I'm not sure about
`SuppressAccessChecks` as turns all kinds of diagnostics off. Did you consider
a possibility to improve `SuppressAccessChecks` to disable only a particular
(specified) type of diagnostics?
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D92024/new/
https://reviews.llvm.org/D92024
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
