[mythtv] [mythtv-commits] mythtv branch master updated bypaulh. v0.28-pre-530-ge02bbc5

[gareth.glaccum at btopenworld.com]

>-----Original Message----- 
>From: Gary Buhrmaster
>On Tue, Nov 19, 2013 at 10:53 PM, Paul Harrison <mythtv at sky.com> wrote:
>.....
>> We should probably make 1.7 the minimum version in that case or just be a
>> devil and jump to 1.8 now? Like Stuart said by the time we release 0.28
>> even
>> dinosaur distros like centos will be using 0.28.
>
>Well, maybe not CentOS 6 (even the enterprise linux vendor's 6.5 beta
>still has taglib 1.6), but by the time of 0.28, CentOS 7 has a fair chance
>to be out.  Since CentOS 6 cannot build the existing MythTV without
>individuals jumping through hoops (due to the Qt 4.8 requirement), I
>would say CentOS 6 compatibility has long ago become something that
>is no longer a requirement (anyone who can build Qt can build a new
>taglib).

Darn, I was just about to respond and say that taglib is a step too far over
having to do QT 4.8 as well.
Oh well.

On my CentOS 6, if there are others which have been 'caught' by this
'incompatibility', I have had the following solutions:
Either
a) revert:
git checkout 45e62c8cb69c3abbbf883d926bd5c0e2137d7365
./configure ....
make ....

or compile up taglib:
install cmake from here:
http://pkgs.repoforge.org/cmake/cmake-2.8.7-1.el6.rfb.x86_64.rpm
(I also installed cppunit-devel doxygen graphviz)
rpmbuild taglib from here:
wget
http://www.mirrorservice.org/sites/dl.fedoraproject.org/pub/fedora/linux/updates/18/SRPMS/taglib-1.8-3.20121215git.fc18.src.rpm

you might need to hack the spec file a bit (it was not a clean install for
me).
rpmbuild --rebuild -ba ./taglib-1.8-3*.src.rpm
If you previously reverted,
git checkout master
then build again.
YMMV


MY master backend appears to be segafaulting a lot now, could be that I have 
a stray slave backend on the network still trying to access it, perhaps 
there is some other problem.

Ultimately, it would be nicer to not have to if the code determine which
taglib is in use. Needing to do the above might reduce the number of people
actively using/testing the code.