=?utf-8?q?Balázs_Kéri?= <[email protected]>
Message-ID:
In-Reply-To: <llvm.org/llvm/llvm-project/pull/[email protected]>


necto wrote:

I'm not against this change, but I wonder if you considered alternatives:

- If you are not afraid of loosing the case of weak+strong definition, a 
simpler approach could be to ignore the duplicates (i.e., in case of a 
duplicate, treat the function as opaque) regardless of weak/strong and not emit 
an error, but maybe a warning or a remark.

- If you do want to keep the case of weak + strong definition, have you 
considered introducing an extra field to defmap, or generate an extra file with 
simple facts about the symbols?
In addition to "weak/strong" (this changeset) they could also, in the future, 
be annotated as `[[noreturn]]`, have their size (for the inline decisions), 
`pure`, etc. If we use an extensible format, we could gradually expand the set 
of properties available to CTU before it decides (or even able) to load the 
external TU.

https://github.com/llvm/llvm-project/pull/204188
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to