[Bug libstdc++/92688] including introduce the name index to global namespace scope
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92688 SODA Noriyuki changed: What|Removed |Added CC||soda-gnu at yuruyuru dot net --- Comment #3 from SODA Noriyuki --- > GCC does not have full control over glibc yes. but glibc have been providing an option to disable the index symbol, and -std=c99, -std=c11 are already using the option. why -std=c++11 won't use the option?
[Bug libstdc++/92688] including introduce the name index to global namespace scope
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92688 --- Comment #6 from SODA Noriyuki --- > More to the point is glibc still does not do the right thing for what is > needed to support C++11 and above: hmm, thanks. the pthread symbols in PR 21327 become visible with -D_XOPEN_SOURCE=700 without -D_GNU_SOURCE. the M_PI symbol in PR 11196 becomes visbile with -D_XOPEN_SOURCE=700 too. is there any other symbols which are necessary for -std=c++11, but invisible with -D_XOPEN_SOURCE=700 ?
[Bug libstdc++/92688] including introduce the name index to global namespace scope
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92688 --- Comment #8 from SODA Noriyuki --- > Libstdc++ cannot define _XOPEN_SOURCE though, > because it could conflict with something the user defines. Yeah, it has similar problem with _GNU_SOURCE, _XOPEN_SOURCE is only closer to what it should be. > The correct fix is for glibc to expose the required names > by some other method (a "backdoor" just for libstdc++) > and not require any feature macros that users should be able to control. I understand. > I still plan to fix this, but it's not a top priority. I hope you'll have some spare time to do so in near future ;-) Thanks for taking you time to explain the issues.
