I have a bug report (https://bugzilla.redhat.com/show_bug.cgi?id=710288) that claims that manaworld 0.0.29 used $RPM_OPT_FLAGS, but that the recent 0.5.2 didn't. Unless there is something subtle it looks like the 0.5.2 is using $RPM_OPT_FLAGS.
One notable difference that I think might have bearing on this is that 0.0.29 used autotools and 0.5.2 uses cmake.
On 06/03/2011 06:19 AM, Bruno Wolff III wrote:
I have a bug report (https://bugzilla.redhat.com/show_bug.cgi?id=710288) that claims that manaworld 0.0.29 used $RPM_OPT_FLAGS, but that the recent 0.5.2 didn't. Unless there is something subtle it looks like the 0.5.2 is using $RPM_OPT_FLAGS.
It obviously doesn't.
C.f. the compiler calls inside of the build.logs, e.g.:
cd /builddir/build/BUILD/tmw-0.5.2/src && /usr/bin/c++ -Wall -I/builddir/build/BUILD/tmw-0.5.2/src -I/usr/include/SDL -I/usr/include/libxml2 -DPACKAGE_VERSION="0.5.2" -DPKG_DATADIR="/usr/share/mana/" -DLOCALEDIR="/usr/share/locale/" -DENABLE_NLS=1 -DUSE_OPENGL -DUSE_X11 -o CMakeFiles/mana.dir/gui/widgets/avatarlistbox.cpp.o -c /builddir/build/BUILD/tmw-0.5.2/src/gui/widgets/avatarlistbox.cpp
If this package was honouring $RPM_OPT_FLAGS, the contents of $RPM_OPT_FLAGS would be found inside of this compiler call.
On Fri, Jun 03, 2011 at 06:46:00 +0200, Ralf Corsepius rc040203@freenet.de wrote:
If this package was honouring $RPM_OPT_FLAGS, the contents of $RPM_OPT_FLAGS would be found inside of this compiler call.
Thanks. I wrongly assumed that setting the CXX flags would result in them getting used. I'll take a deeper look at what is going on.
On Fri, Jun 03, 2011 at 06:27:27 -0500, Bruno Wolff III bruno@wolff.to wrote:
On Fri, Jun 03, 2011 at 06:46:00 +0200, Ralf Corsepius rc040203@freenet.de wrote:
If this package was honouring $RPM_OPT_FLAGS, the contents of $RPM_OPT_FLAGS would be found inside of this compiler call.
Thanks. I wrongly assumed that setting the CXX flags would result in them getting used. I'll take a deeper look at what is going on.
I'll blame it being late when I got home last night and looked at this. For some reason I thought the build information was being hidden when it clearly isn't and just focussed on the CXX flags being set.
Two of the CMakelists.txt files set CMAKE_CXX_FLAGS to -Wall overriding what's passed in. I commented those out and now things appear to look better.
On Fri, 2011-06-03 at 06:59 -0500, Bruno Wolff III wrote:
On Fri, Jun 03, 2011 at 06:27:27 -0500, Bruno Wolff III bruno@wolff.to wrote:
On Fri, Jun 03, 2011 at 06:46:00 +0200, Ralf Corsepius rc040203@freenet.de wrote:
If this package was honouring $RPM_OPT_FLAGS, the contents of $RPM_OPT_FLAGS would be found inside of this compiler call.
Thanks. I wrongly assumed that setting the CXX flags would result in them getting used. I'll take a deeper look at what is going on.
I'll blame it being late when I got home last night and looked at this. For some reason I thought the build information was being hidden when it clearly isn't and just focussed on the CXX flags being set.
Two of the CMakelists.txt files set CMAKE_CXX_FLAGS to -Wall overriding what's passed in. I commented those out and now things appear to look better.
It's always worth sending a patch upstream in this case. I think it should be quite easy, IIRC you can use += in CMake to append instead of overwriting, but check it out, IMBW.