On 26.04.2018 10:17, Peter Maydell wrote: > On 18 April 2018 at 15:44, Daniel P. Berrangé <berra...@redhat.com> wrote: >> On Wed, Apr 18, 2018 at 02:29:47PM +0100, Peter Maydell wrote: >>> Yes, my OSX build system is on glib 2.22. It's using 'fink' for 3rd >>> party library installations, and they somehow seem to have got stuck >>> on 2.22. Looking at the website the project seems to have basically >>> stagnated, and I guess I need to move to homebrew at some point >>> (likely after we get 2.12 out the door). But for the moment this >>> patch would break my pullrequest testing builds. >> >> Yeah that sounds reasonable as it would also align your build env with >> what we are testing via travis too. > > I've now made the switch to homebrew, and so my OSX build tests > will be using glib 2.56. NB that this means that none of my merge > testing will catch "glib version too old" for our current setting > of "too old" -- I'm not sure what my current oldest version is in > the test machine set. > > I've also upgraded Xcode, which means I now have a machine with > a clang that emits all the -Waddress-of-packed-member complaints. > I have configured the test build so those aren't fatal > (via --extra-cflags='-Werror -Wno-error=address-of-packed-member > -Wno-error=deprecated-declarations') but I'm now more interested in > getting those cleaned up so they don't clutter up my logfiles :-)
Isn't the -Wno-error=address-of-packed-members flag added by Gerd's patch already? See: https://git.qemu.org/?p=qemu.git;a=commitdiff;h=0e39c4aa7ecc5699bc391fcb Thomas