On Wed, Dec 16, 2015 at 04:23:44PM +0100, Manfred Hollstein wrote: > > On Wed, Dec 09, 2015 at 07:51:11PM +0100, Manfred Hollstein wrote: > > > FWIW, David's patch is completely right. __USE_* features shouldn't be > > > defined manually, but should be inherited from their proper definition > > > by #include'ing <features.h>, with glibc at least. > > (...) > > > > Thanks for your detailed explanation. Do you suggest that we backport > > it "just in case" ? Nobody seems to have been complaining since 1.5, > > so I'm always a bit worried to risk breaking some obscure system you > > know. > > normally I'd agree with you, but in similar cases like this one, I'd risk > to run into _some_ issues, solely because that code was based on some > wrong assumptions. When we see issues arrising from this change, we > should fix them properly.
Fine, that makes sense. I've backported it to 1.6 and 1.5 then. Cheers, Willy
