As I'm sure you know, MythTV 0.18 was released last night. Please add it to Portage. Thanks! Reproducible: Always Steps to Reproduce: 1. emerge mythtv 2. stare 3. drool Actual Results: mythtv-0.17-r1 installs. Expected Results: mythtv-0.18 installs. Danke.
I suppose we should note that this release might take more ebuild magic since Issac has merged the build into three packages, instead of the previous eleven. This will also affect all of the other official myth ebuilds. It now looks like this: mythplugins - all the official MythTV plugins (MythBrowser, MythDVD, MythGallery, MythGame, MythMusic, MythNews, MythPhone, MythVideo, MythWeather, and MythWeb) myththemes - contains the (large) tarball of myth themes myth - contains the core myth stuffs mythplugins can build individual components by options set, which doesn't appear to be doc'd on the main myth sute, but hopefully is doc'd in the tarball itself.
Created attachment 56513 [details] modified ebuild for mythtv-0.18 Based on a mixture of the mythtv-0.17-r1 ebuild and the mplayer ebuild (and mooch's ebuild in http://forums.gentoo.org/viewtopic-p-2319038.html#2319038), I've fixed the USE flags to the best of my ability. I got the following error on compilation with MMX enabled on P4, celeron and athlon-xp: i686-pc-linux-gnu-gcc -c -pipe -march=pentiumpro -w -O2 -march=athlon-xp -pipe -fno-pie -D_REENTRANT -DPIC -fPIC -DMMX -DUSING_IVTV -DUSING_XRANDR -D_GNU_SOURCE -D_FILE_OFFSET_BITS=64 -DPREFIX=\"/usr\" -DHAVE_AV_CONFIG_H -D_LARGEFILE_SOURCE -DQT_NO_DEBUG -DQT_THREAD_SUPPORT -I/usr/qt/3/mkspecs/linux-g++ -I. -I.. -I../.. -I/usr/qt/3/include -o a52_bitstream.o liba52/a52_bitstream.c libpostproc/postprocess_template.c: In function `postProcess_MMX': libpostproc/postprocess_template.c:3198: error: can't find a register in class `GENERAL_REGS' while reloading `asm' libpostproc/postprocess_template.c:3283: error: can't find a register in class `GENERAL_REGS' while reloading `asm' libpostproc/postprocess_template.c:3198: error: can't find a register in class `GENERAL_REGS' while reloading `asm' libpostproc/postprocess_template.c:3283: error: can't find a register in class `GENERAL_REGS' while reloading `asm' make: *** [postprocess.o] Error 1 make: *** Waiting for unfinished jobs.... make: Leaving directory `/var/tmp/portage/mythtv-0.18/work/mythtv-0.18/libs/libavcodec' Ergo, I've disabled MMX entirely in the ebuild, which obviously isn't ideal - but since it's supposed to be disabled for athlon-xp and AMD64, I can live with it. Hopefully someone comes up with a way to make it work for those platforms which gain an advantage from MMX. I don't think the ebuild actually needs to comment out entries in the settings.pro file any more, but I've converted the modifications from 0.17 for completeness. Hopefully someone with more ebuildfoo than myself can do a better job than this one, but it might prove useful in the meantime.
Created attachment 56524 [details] mythtv-themes-0.18.ebuild Tweaked the mythtv-themes-0.17 ebuild to add the new myththemes tarball, as most of the themes have been moved from mythtv into it.
Created attachment 56527 [details] mythtv-0.18.ebuild Replacing the binary?! version of my mythtv-0.18.ebuild with a text one.
Cardoe's still away, so I'll whip some together and throw them into portage. I'm going to place them in package.mask until all of them are done.
quoting "docs/mythtv-HOWTO.txt": --8<--- As of MythTV version 0.18, there is no need to edit the settings.pro file to set compile-time options. The configure script will configure MythTV appropriately for your system. If you wish to change options, run ./configure --help to see what is available and to override and automatically detected options. See the config.log file after running configure to see previous runs. --8<--- I think we should switch to using configure, this will ensure ebuilds are easily portable to new versions.
Created attachment 56687 [details] MMX enabled mythtv-0.18.ebuild I can successfully compile mythtv with mmx enabled as well as O3 optimizations on both a Pentium4 and an Athlon XP using this modified ebuild.
i686-pc-linux-gnu-gcc -c -pipe -march=pentiumpro -w -O2 -march=athlon-xp -mtune=athlon-xp Why do I have two -march lines with the myth 0.18 ebuild? I think that either mtune and march from make.conf should be ignored if 'above pentiumpro' doesn't work, or drop the -march=pentiumpro. I'm suprised gcc just 'keeps on working'. It compiled and runs fine though. (Except that it's the wrong size, but that might be unrelated : ).
Can use something like: --arch=(filter march value) --extra-cflags="${CFLAGS without march}" --extra-cxxflags="${CXXFLAGS without march}" on configure to pass compile flags. There may be a better way but I have got this to work for my pentium-m and k8 although I hardcoded it.
In portage for three days. Closing.
Um, what about the issues mentioned in comments #8 and #9?
Just to confirm that this issue is still extant. Reopening bug. From the myth-0.18 compile: i686-pc-linux-gnu-gcc -c -pipe -march=pentiumpro -w -march=athlon-xp -O2 -pipe -fno-pie -D_REENTRANT -DPIC -fPIC -DUSING_IVTV -DUSING_XRANDR -DUSING_OPENGL_VSYNC -D_GNU_SOURCE -D_FILE_OFFSET_BITS=64 -DPREFIX=\"/usr\" -DHAVE_AV_CONFIG_H -D_LARGEFILE_SOURCE -DQT_NO_DEBUG -DQT_THREAD_SUPPORT -I/usr/qt/3/mkspecs/linux-g++ -I. -I.. -I../.. -I/usr/qt/3/include -o resample.o liba52/resample.c
Note that #8 doesn't appear to happen in some plugin ebuilds, such as mythweather-0.18, compile snippet shown here: i686-pc-linux-gnu-g++ -c -pipe -Wall -W -march=athlon-xp -O3 -pipe -D_REENTRANT -fPIC -D_GNU_SOURCE -DPREFIX=\"/usr\" -DHAVE_MMX -DQT_NO_DEBUG -DQT_THREAD_SUPPORT -DQT_PLUGIN -I/usr/qt/3/mkspecs/linux-g++ -I. -I/usr/include -I/usr/qt/3/include -o main.o main.cpp i686-pc-linux-gnu-g++ -c -pipe -Wall -W -march=athlon-xp -O3 -pipe -D_REENTRANT -fPIC -D_GNU_SOURCE -DPREFIX=\"/usr\" -DHAVE_MMX -DQT_NO_DEBUG -DQT_THREAD_SUPPORT -DQT_PLUGIN -I/usr/qt/3/mkspecs/linux-g++ -I. -I/usr/include -I/usr/qt/3/include -o weather.o weather.cpp /usr/qt/3/bin/moc weather.h -o moc_weather.cpp
If you have problems with the ebuild _in portage_ (if it does not work for you), file a bug about that ebuild. This bug is about version bump which has been done --> FIXED.
