[mythtv] rpm build of master fails after ffmpeg refactor
John Pilkington
johnpilk222 at gmail.com
Thu May 10 20:12:15 UTC 2018
On 10/05/18 20:19, Peter Bennett wrote:
>
>
> On 05/10/2018 10:07 AM, John Pilkington wrote:
>> I just realised (after a comment by Paul H) that commits to master
>> hadn't stopped but are not coming through on the mailing list, so this
>> is a bit late.
>>
>> I'm using a specfile based on one from rpmfusion. Last success was on
>> 25 April. Now I see
>>
>> + ./configure --qmake=/usr/bin/qmake-qt5 --prefix=/usr
>> --libdir=/usr/lib64 --libdir-name=lib64 --mandir=/usr/share/man
>> --enable-libmp3lame --enable-libtheora --enable-libvorbis
>> --enable-libx264 --enable-libxvid ' '
>> Unknown option "--enable-libtheora".
>>
>> All those 'enables' come from a section of the specfile with this
>> preamble:
>>
>> %build
>>
>> # First, we build MythTV
>> pushd mythtv
>>
>> # Similar to 'percent' configure, but without {_target_platform} and
>> # {_exec_prefix} etc... MythTV no longer accepts the parameters that the
>> # configure macro passes, so we do this manually.
>> ./configure \
>> --qmake=%{_bindir}/qmake-qt5 \
>> --prefix=%{_prefix} \
>> --libdir=%{_libdir} \
>> --libdir-name=%{_lib} \
>> --mandir=%{_mandir} \
>> <snip>
>> --enable-bdjava \
>> --enable-libmp3lame \
>> --enable-libtheora --enable-libvorbis \
>> --enable-libx264 \
>> --enable-libx265 \
>> --enable-libxvid \
>> --enable-libvpx \
>>
>> and I suspect that the others would fail too. I tried
>> --enable-libmyththeora, but no joy.
>>
>> This is pretty early in the build process and I have no doubt that
>> fixing it will simply reveal the need for lots more changes; but
>> perhaps someone will be able to suggest an approach?
>>
>> And do I need to disable plugins?
>>
>> John P
>>
>> gmail doesn't show me my own posts so this is 'edited as new'
>>
>> After a few more attempts I deleted everything after the <snip> above
>> until the following 'make' line, and building progressed to the
>> plugins, when it eventually reported 'no rule to make
>> target...libmyth-30.so' needed by 3 components of mytharchive, and
>> quit. Reported build time was 42 minutes, rather than 8 minutes as
>> before. git was f7ff74a and so includes the install-mytharchive
>> without-plugins commit at dca00e0
>>
>> I'm minded to try a build --without-plugins; the only one I have used
>> often is mytharchive.
>>
>> -----------
>> I downloaded the latest 6e939ad and tried as above but --without plugins
>>
>> This failed (45 min) in mythtv-common in the 'files' section:
>>
>> Processing files: mythtv-common-master-30.630.a.fc26.x86_64
>> error: File not found:
>> /builddir/build/BUILDROOT/mythtv-master-30.630.a.fc26.x86_64/usr/bin/mythwikiscripts
>>
>>
>> I see no other references to mythwikiscripts in the specfile. My
>> earlier builds do include it.
>>
>> That folder does contain several other files:
>>
>> $ ls
>> /var/lib/mock/fedora-26-x86_64/root/builddir/build/BUILDROOT/mythtv-master-30.630.a.fc26.x86_64/usr/bin
>>
>>
>> mythavtest mythcommflag mythffserver mythfrontend
>> mythlcdserver mythpreviewgen mythshutdown mythtv-setup
>> optimize_mythdb
>> mythbackend mythffmpeg mythfilerecorder mythhdhomerun_config
>> mythmediaserver mythreplex mythtranscode mythutil
>> mythccextractor mythffprobe mythfilldatabase mythjobqueue
>> mythmetadatalookup mythscreenwizard mythtvsetup mythwelcome
>>
>> The build log has no references to mytharchive.
>> -------------
>>
>>
>>
> I don't know why you are getting so many strange errors. Was this
> working before? For example, a normal build of MythTV does not put
> mythwikiscripts in /usr/bin . The plugins build does put it in
> /usr/local/bin.
>
> I think your "without-plugins" is still trying to build plugins.
>
> I did not change anything in the plugins configure. There were android
> related changes to the plugins configure. Perhaps those caused a problem.
>
> Peter
I'm not particularly surprised by what I'm seeing. It's been clear for
a long time that the rpm builds aren't the same as those used in the
buildbots - they have had different evolutionary paths, and I have
(almost) always used the rpms. Both have been made to work, and will
normally stay working, but changes often do affect them differently.
John
More information about the mythtv-dev
mailing list