rnk wrote:

Thanks! Yes, I think our ideal user experience here is actually to point back 
at the original user `#include <intrin.h>` and issue a header-not-found error. 
That's the purpose of the include_next, to basically fall back to whatever user 
include might exist. I felt that squatting the global "intrin.h" header name on 
non-Windows platforms was a bit aggressive.

With that in mind, can this break user code, if someone is using `#include 
"intrin.h"` on Linux? Is there a way to do a not-insanely-expensive search of 
the corpus of C++ code on github for files named intrin.h which are not the 
MSVC or Clang builtin headers to gauge the risk? If it came up empty, then 
maybe we turn this into a hard error altogether for non-`_WIN32` builds and add 
a release note.

https://github.com/llvm/llvm-project/pull/234114
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to