GNOME Bugzilla – Bug 536780
gcc 4.3.0 compile warnings
Last modified: 2008-06-09 06:29:57 UTC
Current core fails to build with the following: cc1: warnings being treated as errors gstobject.c: In function 'gst_object_set_name_default': gstobject.c:606: error: dereferencing type-punned pointer will break strict-aliasing rules gstobject.c:616: error: dereferencing type-punned pointer will break strict-aliasing rules
<Uraeus> mclasen: any guess for a timeline for solving the glib2 - dereferencing type-punned pointer will break strict-aliasing rules issue in fedora? --> jaegerflex (~felix@e181057133.adsl.alicedsl.de) has joined #gnome-hackers <-- pochu has quit (leaving) --> Benoit (~bdejean@lns-bzn-48f-81-56-180-111.adsl.proxad.net) has joined #gnome-hackers <mclasen> Uraeus: I don't know that it can be fixed <Uraeus> mclasen: hmm, that would kill the no-warnings policy of gstreamer :( <mclasen> Uraeus: no-warnings policies are pretty much doomed to failure <Uraeus> mclasen: we been able to abide by it for 3-4 years now :) <jdahlin> Uraeus: gstreamer can always disable that particular warning. <jdahlin> -fno-strict-aliasing iirc <slomo__> Uraeus: http://bugzilla.gnome.org/show_bug.cgi?id=316221 <--- take that patch, should make it warning clean unless you're on ppc or some other weird architecture ;) <bugbot> Bug 316221: normal, Normal, ---, gtkdev@gtk.org, NEW, G_LOCK warns about breaking strict-aliasing <slomo__> but the other warning should definitely be fixed somehow no idea how :) <Uraeus> slomo__: I do not want to patch my glib, cause I would have to redo that every time mclasen and redhat updates the system packages. So if that patch isn't something mclasen wants to merge I assume its because its not good enough?
2008-06-05 Tim-Philipp Müller <tim.muller at collabora co uk> * m4/gst-error.m4: Add -fno-strict-aliasing when compiling with -Werror, to work around warnings caused by G_LOCK with recent GLib versions (2.16.x) (#316221).