Hi, On Sun, Dec 20, 2015 at 11:17:03AM +0100, Samuel Thibault wrote: > Samuel Thibault, on Fri 18 Sep 2015 09:20:39 +0200, wrote: > > merely putting the definition of _GNU_SOURCE before #includes (as it > > should always be) fixes the build, here is a tested patch. > > Ping?
Thank you for the reminder. Upstream don't want to carry the patch as they can't test it, so I was about to commit it. But then I noticed that _GNU_SOURCE is alreadying being set in the current "hurd.patch" quilt patch. Is the current patch no longer working? Can it be updated or removed? Could someone please provide a patch that updates everything cleanly so that existing overlapping Hurd-related patches don't bit rot? I would prefer if possible to ensure that when built on GNU/Linux we're building as close as possible to what upstream shipped, since that's what they test. Given that upstream won't take patches for GNU/Hurd we can carry a GNU/Hurd patch if someone can maintain that, but it should be clear to not affect the build on GNU/Linux in that case. I'd expect something like preprocessor conditionals or (like the current hurd.patch does) a conditional adjustment to the configure stage to achieve this. Sorry for the delay in telling you this now. I had been focused on things like replacing mysql-5.5 in testing before. Robie
signature.asc
Description: Digital signature
