Re: [lmi] LLVM libc++

[Greg Chicares]

On 10/5/22 17:38, Vadim Zeitlin wrote:
> On Wed, 5 Oct 2022 16:45:48 +0000 Greg Chicares <gchicares@sbcglobal.net> 
> wrote:
> 
> GC> This problem is observed in the presence of llvm libunwind,
> GC> but not observed in its absence (i.e., with HP libunwind).
> GC> That's all I know.
> 
>  Please let me know if you'd like me to try to reproduce this and
> understand what happens there. I won't do it immediately because, perhaps,
> it's not really necessary if using pragmas provides a satisfactory
> solution/workaround.

My casual opinion is that it's unnecessary. The libstdc++ <regex>
library apparently is not UBSAN-clean, or UBSAN just doesn't work
perfectly with gcc; either way, the presence or absence of error
messages may be aligned with random circumstances beyond our
control.

And using a pragma does work. You could see that in the two small
commits that I just attempted to push, except for this problem:
  sr #110737: git push rejected
  https://savannah.nongnu.org/support/index.php?110737