Hi, Bill, Thank you for the info.
> On Aug 21, 2024, at 17:36, Bill Wendling <isanb...@gmail.com> wrote: > >> >> Bill, could you please provide a little bit more info on the possibility of >> a new builtin __builtin_has_attribute() in CLANG? >> > From what I gathered, it would require some moderate surgery to the > Clang front-end to support "__builtin_has_attribute()". Parsing of > attributes is done only in the context of the "__attribute__" keyword > and would have to be refactored. That in itself would probably be a > nice clean-up to the Clang front-end. The bigger issue is how to > support an attribute as a node in the AST. We don't currently have a > node like that in our AST, because attributes are associated with an > AST node and aren't nodes in and of themselves. So it's a lot more > work than it appears on the surface. > > That said, Clang probably *should* support __builtin_has_attribute(). > The question is whether we should do it for this feature or wait? If > we do it for this feature, I imagine that Clang's implementation of > __builtin_get_counted_by() will be significantly delayed, but if it's > the correct thing to do then we should bite the bullet. >From the discussion so far, I think that __builtin_get_counted_by () might be >good enough. Let me know if there is any objection on this. Qing > > -bw >
