Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!
Bug 198849 - xbmc ebuild request
Summary: xbmc ebuild request
Status: RESOLVED FIXED
Alias: None
Product: Gentoo Linux
Classification: Unclassified
Component: New packages (show other bugs)
Hardware: All Linux
: High enhancement with 6 votes (vote)
Assignee: Default Assignee for New Packages
URL: http://www.xboxmediacenter.com/
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2007-11-11 19:50 UTC by Jesus de Santos Garcia
Modified: 2009-02-24 23:38 UTC (History)
31 users (show)

See Also:
Package list:
Runtime testing required: ---


Attachments
Basic XBMC SVN ebuild (xbmc-9999.ebuild,1.34 KB, text/plain)
2008-05-31 08:42 UTC, BT
Details
Improved ebuild (xbmc-20080609.ebuild,1.56 KB, text/plain)
2008-06-12 14:12 UTC, Daniel Burr
Details
Improved ebuild (xbmc-20080625.ebuild,1.72 KB, text/plain)
2008-06-26 00:02 UTC, Daniel Burr
Details
Work around for "Floating Point Exception" (alsa-buffer-size-fix.patch,792 bytes, patch)
2008-06-26 00:04 UTC, Daniel Burr
Details | Diff
Improved Ebuild (xbmc-20080702.ebuild,1.63 KB, text/plain)
2008-07-02 18:08 UTC, Gustaf Ankarloo
Details
New Ebuild (xbmc-20080707.ebuild,2.36 KB, text/plain)
2008-07-07 12:41 UTC, Gustaf Ankarloo
Details
Updated ebuild (xbmc-9999.ebuild,1.97 KB, text/plain)
2008-07-10 12:33 UTC, BT
Details
Updated ebuild (xbmc-9999.ebuild,1.99 KB, text/plain)
2008-07-22 06:36 UTC, Marco Genasci
Details
Updated ebuild (xbmc-9999.ebuild,2.05 KB, text/plain)
2008-07-22 12:18 UTC, BT
Details
Updated ebuild (xbmc-9999.ebuild,2.04 KB, text/plain)
2008-07-27 09:34 UTC, BT
Details
Updated ebuild (xbmc-9999.ebuild,2.05 KB, text/plain)
2008-07-28 04:45 UTC, BT
Details
Updated ebuild (xbmc-9999.ebuild,2.25 KB, text/plain)
2008-08-03 06:14 UTC, BT
Details
Beta 1 ebuild (xbmc-8.10_beta1.ebuild,2.25 KB, text/plain)
2008-09-19 12:45 UTC, BT
Details
Pegged ebuild (xbmc-8.10_p15828.ebuild,3.76 KB, text/plain)
2008-10-09 16:31 UTC, Peter Fern
Details
Trunk version of previous ebuild (xbmc-9999.ebuild,3.65 KB, text/plain)
2008-10-09 16:33 UTC, Peter Fern
Details
Patch to remove some XMMS tests (xmms-deprecated.patch,1.15 KB, patch)
2008-10-09 16:36 UTC, Peter Fern
Details | Diff
Pegged ebuild (xbmc-8.10_p15940.ebuild,3.96 KB, text/plain)
2008-10-20 06:50 UTC, Peter Fern
Details
Trunk version of previous ebuild (xbmc-9999.ebuild,3.77 KB, text/plain)
2008-10-20 06:51 UTC, Peter Fern
Details
xbmc-9999-r1.ebuild (xbmc-9999-r1.ebuild,6.59 KB, text/plain)
2008-10-28 04:33 UTC, Rick Harris
Details
cutout of build.log for sandbox issue (build.log,15.32 KB, text/plain)
2008-11-03 17:53 UTC, Per Öberg
Details
xbmc-9999-r2.ebuild (xbmc-9999-r2.ebuild,5.73 KB, text/plain)
2008-11-05 22:57 UTC, Rick Harris
Details
xbmc-9999-r3.ebuild (xbmc-9999-r3.ebuild,6.29 KB, text/plain)
2008-11-07 12:03 UTC, Rick Harris
Details
xbmc-9999-r4.ebuild (xbmc-9999-r4.ebuild,7.71 KB, text/plain)
2008-11-12 07:20 UTC, Rick Harris
Details
xbmc-9999-r5.ebuild (xbmc-9999-r5.ebuild,7.32 KB, text/plain)
2008-11-13 11:44 UTC, Rick Harris
Details
xbmc-plugininstaller-9999.ebuild (xbmc-plugininstaller-9999.ebuild,1.88 KB, text/plain)
2008-11-15 10:45 UTC, Rick Harris
Details
xbmc-8.10.ebuild (xbmc-8.10.ebuild,8.78 KB, text/plain)
2008-11-16 11:32 UTC, Rick Harris
Details
Updated ebuild for xbmc-8.10 (xbmc-8.10-r1.ebuild,8.75 KB, text/plain)
2008-11-25 12:08 UTC, Daniel Plaza
Details
xbmc-9999.ebuild (xbmc-9999.ebuild,8.85 KB, text/plain)
2008-12-05 20:58 UTC, Rick Harris
Details
xbmc-8.10-r2.ebuild (xbmc-8.10-r2.ebuild,8.88 KB, text/plain)
2008-12-12 03:08 UTC, Rick Harris
Details
xbmc-8.10-r2.ebuild (xbmc-8.10-r2.ebuild,8.90 KB, text/plain)
2008-12-12 08:01 UTC, Rick Harris
Details
xbmc-8.10-r3.ebuild (xbmc-8.10-r3.ebuild,9.08 KB, text/plain)
2008-12-15 10:57 UTC, Rick Harris
Details
xbmc-8.10-r4.ebuild (xbmc-8.10-r4.ebuild,9.23 KB, text/plain)
2008-12-25 09:22 UTC, Rick Harris
Details
xbmc.readsector.patch (xbmc.readsector.patch,2.85 KB, text/plain)
2008-12-25 09:23 UTC, Rick Harris
Details
xbmc-9999-r1.ebuild (xbmc-9999-r1.ebuild,8.99 KB, text/plain)
2008-12-25 09:28 UTC, Rick Harris
Details
xbmc-plugininstaller-9999-r1.ebuild (xbmc-plugininstaller-9999-r1.ebuild,1.97 KB, text/plain)
2008-12-27 20:59 UTC, Rick Harris
Details
log of failing install (build.log,1.32 MB, text/plain)
2008-12-31 14:02 UTC, alex
Details
xbmc-8.10-r5.ebuild (xbmc-8.10-r5.ebuild,9.24 KB, text/plain)
2009-01-04 23:53 UTC, Rick Harris
Details
xbmc-9999-r2.ebuild (xbmc-9999-r2.ebuild,9.00 KB, text/plain)
2009-01-04 23:53 UTC, Rick Harris
Details
live ebuild with additional USE flags (xbmc-9999-r3.ebuild,7.89 KB, text/plain)
2009-02-19 02:16 UTC, Peter Fern
Details
Added warnings for flags tracking alternate branches (xbmc-9999-r3.ebuild,8.24 KB, text/plain)
2009-02-19 02:24 UTC, Peter Fern
Details
renamed patch (xbmc-9999_readsector.patch,2.85 KB, text/plain)
2009-02-19 22:52 UTC, Peter Fern
Details
Added some eautoreconfs (xbmc-9999-r3.ebuild,8.31 KB, text/plain)
2009-02-20 21:39 UTC, Peter Fern
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Jesus de Santos Garcia 2007-11-11 19:50:30 UTC
XBMC is being ported to Linux. I wonder if we could start a ebuild to test the evolution of this great package.

I can volunteer to create the first version but I am not sure how to do this when there is no tar.gz available (only svn).
Comment 1 Jakub Moc (RETIRED) gentoo-dev 2007-11-11 21:37:07 UTC
So attach one. Other than that, there's just 2500+ ebuilds waiting for inclusion...
Comment 2 Barry Gear 2008-05-15 16:05:13 UTC
README is here (includes build instructions): http://xbmc.svn.sourceforge.net/viewvc/*checkout*/xbmc/branches/linuxport/XBMC/README.linux

I might try to make one this weekend..
Comment 3 Nicholas Bates 2008-05-27 08:09:34 UTC
(In reply to comment #2)
> README is here (includes build instructions):
> http://xbmc.svn.sourceforge.net/viewvc/*checkout*/xbmc/branches/linuxport/XBMC/README.linux
> 
> I might try to make one this weekend..
> 

How did you go with this Barry?
Comment 4 BT 2008-05-31 08:42:44 UTC
Created attachment 154915 [details]
Basic XBMC SVN ebuild

I have posted a basic XBMC SVN ebuild which essentially just pulls in the required dependencies. I'm not sure if it still works as I currently don't have access to a Gentoo test machine, but it's a good starting point for anyone wanting to develop the ebuild further.
Comment 5 Daniel Burr 2008-06-12 14:12:56 UTC
Created attachment 156499 [details]
Improved ebuild

A few changes:
* Don't need samba as a dependency because xbmc has it's own internal version
* OpenGL support if opengl USE flag is specified
* Use einstall instead of emake to fix sandbox violations

With this new ebuild I was able to emerge XBMC successfully!
Comment 6 BT 2008-06-13 01:52:26 UTC
(In reply to comment #5)
> With this new ebuild I was able to emerge XBMC successfully!

Great work on fixing the sandbox and dependency issues. The original ebuild was also missing the sdl-stretch dependency listed in README.linux. Not sure if it's required, as XBMC appears to build without it.
Comment 7 Daniel Burr 2008-06-26 00:02:45 UTC
Created attachment 158461 [details]
Improved ebuild

XBMC SVN now requires boost, added it.  Also added a work-around for the "floating point exception".
Comment 8 Daniel Burr 2008-06-26 00:04:07 UTC
Created attachment 158463 [details, diff]
Work around for "Floating Point Exception"

See http://xbmc.org/forum/showthread.php?t=29076
Comment 9 Gustaf Ankarloo 2008-07-02 10:52:09 UTC
The workaround don't work for me with Intel HD Audio Alsa driver.
only solution is to emerge alsa-lib with debug useflag
Comment 10 Gustaf Ankarloo 2008-07-02 13:42:40 UTC
Also there would be nice if halmount was optional. 
my system dont like that at all...but without it...everything just dandy.

If I knew how I would make the changes my self.
Comment 11 BT 2008-07-02 13:52:57 UTC
I believe halmount is optional and is controlled via the hal use flag. Have you tried disabling it?
Comment 12 Gustaf Ankarloo 2008-07-02 18:08:31 UTC
Created attachment 159349 [details]
Improved Ebuild

Updated Ebuild.
cleaner and more useflags
Comment 13 Gustaf Ankarloo 2008-07-02 18:13:20 UTC
My bad BT.
I don't know what went trough my mind....of course tha hal useflag disables halmount. 

And I removed the patch from Ebuild...it dont work on my system and Iäve read that it works for some but emerging alsa-lib with debug seems to solve it for all.
Think thats the best way to go.
Comment 14 Gustaf Ankarloo 2008-07-07 12:41:16 UTC
Created attachment 159791 [details]
New Ebuild

Added information how to use ESVN_REVISION to either get the latest revision or a specific.
Comment 15 Trent Nelson 2008-07-08 18:46:31 UTC
Hey guys,

   Great job taking the initiative to put together an ebuild for XBMC.  I'd like to invite you all to join our forums (if you haven't already) at http://xbmc.org/forum and advertise your work.  I know we have several Gentoo users (myself being one), so you will likely get all kinds of "fun" questions and bug reports/feature requests. This is, of course, merely an invitation and may be declined if you aren't interested.
   Also, (I don't speak .ebuild so I'm going off observation and assumption) there was a recent backtrace posted in the forums without debug symbols.  The user who posted it stated their install method was "manual/ebuild" so presumably they used this ebuild (I know of no others).  I was wondering if you guys could force USE=debug at least until we make a release that we feel is stable.  This would greatly assist in debugging efforts.

Thanks,
ATK
Comment 16 BT 2008-07-10 12:33:08 UTC
Created attachment 160046 [details]
Updated ebuild

Changes include:
- Added fontconfig, unrar and zip dependencies.
- Removed the libsdl dependency as it is already being pulled in by the required media-libs/sdl-* dependencies.
- Check if libsdl has not been built with USE nojoystick when joystick USE is set.
- Filter --fomit-frame-pointer flag when gprof USE is set.
- Removed the ALSA patch by Daniel, as it doesn't seem to work.
- Forced --enable-debug at the request of upstream until stable release.
Comment 17 Marco Genasci 2008-07-22 06:36:16 UTC
Created attachment 161074 [details]
Updated ebuild

Added cmake dependencies.
Comment 18 BT 2008-07-22 12:18:29 UTC
Created attachment 161090 [details]
Updated ebuild

Changes include:
- Added nasm, faac, enca, cmake, libXt, libXmu and gperf dependencies.
- Removed libmms dependency.
Comment 19 BT 2008-07-27 09:34:52 UTC
Created attachment 161453 [details]
Updated ebuild

Changes include:
- Added ESVN_REVISION variable. It lets you control which SVN revision you want to build. It is set to HEAD by default. Thanks to Gustaf for the idea :)
- Added unzip dependency.
- Removed python, unrar and zip dependencies.
Comment 20 BT 2008-07-28 04:45:18 UTC
Created attachment 161515 [details]
Updated ebuild

Changes include:
- Added zip dependency which was mistakenly removed.
Comment 21 BT 2008-08-03 06:14:30 UTC
Created attachment 162061 [details]
Updated ebuild

Changes include:
- Added ~amd64 keyword (Thanks to TehUni for testing).
- Added yasm dependency for amd64.
- Added jpeg and libpng dependencies.
- Renamed hal USE to automount.
- Check if pmount has been built with USE hal when automount USE is
set.
- Removed the hal dependency as it will be pulled in by pmount when automount USE is set.
Comment 22 Kelvie Wong 2008-08-24 09:40:51 UTC
(In reply to comment #21)
> Created an attachment (id=162061) [edit]
> Updated ebuild
> 

I'm getting libtool / ltmain.sh errors for a lot of the subprojects inside the XBMC code tree (such as all of the dvd codec libraries and goom)

Message:

*** [Gentoo] sanity check failed! ***
*** libtool.m4 and ltmain.sh have a version mismatch! ***
*** (libtool.m4 = 1.5.24, ltmain.sh = "1.5.26 Debian 1.5.26-1ubuntu1") ***

Please run:

  libtoolize --copy --force


--
I do run that in each respective subdirectory, and do a FEATURES=keepwork emerge --resume

and the compile seems to keep going.
Comment 23 BT 2008-08-25 06:32:20 UTC
(In reply to comment #22)
> I do run that in each respective subdirectory, and do a FEATURES=keepwork
> emerge --resume
> 
> and the compile seems to keep going.

Could you please add elibtoolize to src_unpack() right after subversion_src_unpack and see if that fixes the issue for you. I'll post an updated ebuild if it works.
Comment 24 Omega Xi 2008-08-28 04:23:45 UTC
(In reply to comment #21)

Sorry to be a pain, most of this is new to me, so I'm not sure if I made a mistake somewhere.  I tried to use this ebuild (attachment 162061 [details]), however when I tried to emerge xbmc I get the following message

 * 
 * ERROR: media/xbmc-9999 failed.
 * Call stack:
 *               ebuild.sh, line   49:  Called src_compile
 *             environment, line 2429:  Called die
 * The specific snippet of code:
 *       emake || die "Make failed!"
 *  The die message:
 *   Make failed!
 * 

I have put the contents of /var/tmp/portage/media/xbmc-9999/temp/build.log into a file along with the output from "emerge --info" you can find it here:
http://www.omega-xi.com/log/xbmc-emerge.log

I'm really sorry if that isn't very helpful or if the reason it didn't compile is simply because I have made a mistake, I am providing this information in the hope it will be usefull to you not as a request for support.
Comment 25 BT 2008-08-28 05:11:27 UTC
(In reply to comment #24)

If you take a look at the log you should notice an error when compiling the Goom visualization. It appears to be one of many random XBMC build failures. I think these failures are caused by the fact the many of the XBMC Makefiles are passing CFLAGS=-O3 which from my understanding is not supported by Gentoo.

You can try and emerge XBMC again as it might build the next time round. I plan to post an ebuild very soon which uses the tagged alpha 4 release of XBMC. This will avoid the SVN breakages which occur too often with the 9999 ebuild. I will also need to filter out or sed the O3's from the Makefiles.

Thanks for the feedback.
Comment 26 Dennis Oberhoff 2008-09-08 14:20:52 UTC
I'm not sure why, but the ebuild fails for me at building libPython. This problem does not appear using the build.sh script.

/var/tmp/portage/media-video/xbmc-20090809/work/xbmc-20090809/xbmc/lib/libPython/Python/Lib /var/tmp/portage/media-video/xbmc-20090809/work/xbmc-20090809/xbmc/lib/libPython/linux
/var/tmp/portage/media-video/xbmc-20090809/work/xbmc-20090809/xbmc/lib/libPython/linux
make[2]: Leaving directory `/var/tmp/portage/media-video/xbmc-20090809/work/xbmc-20090809/xbmc/lib/libPython/linux'
make[1]: Leaving directory `/var/tmp/portage/media-video/xbmc-20090809/work/xbmc-20090809/xbmc/lib/libPython/linux'
 * 
 * ERROR: media-video/xbmc-20090809 failed.
 * Call stack:
 *               ebuild.sh, line   49:  Called src_compile
 *             environment, line 2421:  Called die
 * The specific snippet of code:
 *       emake || die "Make failed!"
 *  The die message:
 *   Make failed!
 * 
 * If you need support, post the topmost build error, and the call stack if relevant.
 * A complete build log is located at '/var/tmp/portage/media-video/xbmc-20090809/temp/build.log'.
 * The ebuild environment file is located at '/var/tmp/portage/media-video/xbmc-20090809/temp/environment'.
 * This ebuild is from an overlay: '/usr/local/mythtv-overlay/'
 * 


Comment 27 Per Wigren 2008-09-18 17:32:56 UTC
It fails for me also. It looks like it fails when building libPython but because the make is parallell (-j3) it fails a bit earlier but the remaining jobs continue to compile until they are done.

The error in the log file is this:

configure.in: required file `./ylwrap' not found
configure.in:   `automake --add-missing' can install `ylwrap'
make[2]: *** [Makefile.in] Error 1
make[2]: Leaving directory `/mnt/archive/temp/sys/vartmp/portage/media-tv/xbmc-9999/work/xbmc-9999/xbmc/visualizations/Goom/goom2k4-0'
make[1]: *** [goom2k4-0/src/.libs/libgoom2.a] Error 2
make[1]: Leaving directory `/mnt/archive/temp/sys/vartmp/portage/media-tv/xbmc-9999/work/xbmc-9999/xbmc/visualizations/Goom'
make: *** [visualizations] Error 2
Comment 28 BT 2008-09-19 12:45:39 UTC
Created attachment 165834 [details]
Beta 1 ebuild

Changes include:
- Added eautoreconf in src_unpack().
- ESVN_REPO_URI now points to the tagged beta 1 release.

I have successfully built the beta 1 release numerous times without any of the reported build failures. It should be much more stable since it's a tagged release, although the ebuild still uses Subversion since there are no tarballs. I recommend that users no longer use the 9999 ebuild as SVN trunk breaks way too often.
Comment 29 Mathy Vanvoorden 2008-09-21 14:06:39 UTC
I can confirm that the ebuild works on amd64.
I also ran into the "Floating Point Exception"-bug that is fixed by adding debug to alsa-lib USE-flags.
Comment 30 Mathy Vanvoorden 2008-09-22 20:16:13 UTC
I have also tried this ebuild on an i686 box (my actual htpc) and experienced there the same problem as described by Kelvie. I tried the elibtoolize suggestion but that had no succes. I am currently trying to build it by manually fixing the problem and then resuming the installation, but it is still compiling so I do not yet know if this method will work for me.
Comment 31 BT 2008-09-23 00:27:24 UTC
(In reply to comment #30)
> I have also tried this ebuild on an i686 box (my actual htpc) and experienced
> there the same problem as described by Kelvie.

I haven't been able to replicate this or the other build errors. It appears that it only happens on x86. Can anyone replicate them on amd64?

> I tried the elibtoolize suggestion but that had no succes.

The beta 1 ebuild runs eautoreconf which calls elibtoolize so that suggestion is no longer valid.
Comment 32 Archimedes Trajano 2008-09-23 07:01:07 UTC
Can it be done so instead of one ebuild that does everything we have multiple ebuilds to make it more modular?
Comment 33 BT 2008-09-23 07:18:38 UTC
(In reply to comment #32)
> Can it be done so instead of one ebuild that does everything we have multiple
> ebuilds to make it more modular?

I don't believe this is possible.
Comment 34 Gustaf Ankarloo 2008-09-23 09:30:46 UTC
(In reply to comment #32)
> Can it be done so instead of one ebuild that does everything we have multiple
> ebuilds to make it more modular?

What do you mean by more modular?
isn't the USE flags enough.
Comment 35 Xake 2008-09-25 10:22:49 UTC
(In reply to comment #32)
> Can it be done so instead of one ebuild that does everything we have multiple
> ebuilds to make it more modular?
> 

Sorry, but this thing is not built currently to be modular. If it really was, then you would be able to use the system ffmpeg and so on instead of the by SVN provided one.
But that is something that should be carried upstream, not in the ebuild.
Comment 36 Per Wigren 2008-09-25 13:29:11 UTC
I tried the new beta1 ebuild, Now it fails like this instead:

checking for dld_link in -ldld... Compiling ../Python/Lib/ntpath.pyc
no
./configure: line 7995: _LT_CMD_GLOBAL_SYMBOLS: command not found
checking for _ prefix in compiled symbols... Compiling ../Python/Lib/nturl2path.pyc
no
./configure: line 8054: syntax error near unexpected token `newline'
./configure: line 8054: `         _LT_TRY_DLOPEN_SELF('
make[2]: *** [config.status] Error 2
make[2]: Leaving directory `/mnt/archive/temp/sys/vartmp/portage/media-tv/xbmc-8.10_beta1/work/xbmc-8.10_beta1/xbmc/screensavers/rsxs-0.9'
make[1]: *** [..] Error 2
make[1]: Leaving directory `/mnt/archive/temp/sys/vartmp/portage/media-tv/xbmc-8.10_beta1/work/xbmc-8.10_beta1/xbmc/screensavers/rsxs-0.9/xbmc'
make: *** [screensavers] Error 2

I'm on amd64 with ACCEPT_KEYWORDS="~amd64".

automake/autoconf/libtool version mismatch?
Comment 37 Daniel Marmander 2008-10-02 19:26:25 UTC
It compiles fine, but no pointer shows and some graphics seems to be missing (nothing shows which menu is selected). According to the xbmc forum, this does not occur when compiling manually, not using the ebuild.

Trying to re-install it results in errors, unless the traces of the first installation is removed (ie the source code).

The problem is described in: http://xbmc.org/forum/showthread.php?t=36795
Comment 38 Peter Fern 2008-10-09 16:31:19 UTC
Created attachment 167750 [details]
Pegged ebuild

This is an adaptation of the last ebuild to emulate some of the build.sh actions, and to overcome the Gentoo autotools sanity checks introduced recently.  This version is pegged to a revision (just change the _p# to match the rev you're after).

Next step for me will be to look at splitting out the skins and web themes, then adding a XBMC user and autostart setup.

Please note that this ebuild has not been tested to produce a functional build (I only have headless Gentoo boxen immediately), however it does compile and install correctly.
Comment 39 Peter Fern 2008-10-09 16:33:35 UTC
Created attachment 167752 [details]
Trunk version of previous ebuild

This is the trunk version of the previous ebuild (un-keyworded).

Again, please note that this ebuild has not been tested to produce a functional build.
Comment 40 Peter Fern 2008-10-09 16:36:21 UTC
Created attachment 167754 [details, diff]
Patch to remove some XMMS tests

This patch is for the previous two ebuilds, and removes some tests for XMMS since it's no longer available in portage, and automake barfs..

As a final note, feel free to tell me I'm smoking crack with all the eautoreconf stuff in these ebuilds, seemed the easiest option to me though.
Comment 41 Peter Fox 2008-10-10 19:43:31 UTC
I had a go at building the xbmc-9999 ebuild, and it dies due to a latex error processing the libdvdcss documentation:

Chapter 3.
(./dvdcss_8h.tex [5]

! Package inputenc Error: Unicode char \u8:éph not set up for use with LaTeX.

See the inputenc package documentation for explanation.
Type  H <return>  for immediate help.
 ...

l.49 \item[Author:]Stéph
                        ane Borel $<${\tt stef@via.ecp.fr}$>$
?
! Emergency stop.
 ...

l.49 \item[Author:]Stéph
                        ane Borel $<${\tt stef@via.ecp.fr}$>$
Output written on refman.dvi (9 pages, 8540 bytes).
Transcript written on refman.log.
make[5]: *** [refman.dvi] Error 1
make[5]: Leaving directory `/var/tmp/portage/media-video/xbmc-9999/work/xbmc-9999/xbmc/cores/dvdplayer/Codecs/libDVDCSS/doc/\
latex'
make[4]: *** [stamp-latex] Error 2
make[4]: Leaving directory `/var/tmp/portage/media-video/xbmc-9999/work/xbmc-9999/xbmc/cores/dvdplayer/Codecs/libDVDCSS/doc'
make[3]: *** [all-recursive] Error 1
make[3]: Leaving directory `/var/tmp/portage/media-video/xbmc-9999/work/xbmc-9999/xbmc/cores/dvdplayer/Codecs/libDVDCSS'
make[2]: *** [all] Error 2
make[2]: Leaving directory `/var/tmp/portage/media-video/xbmc-9999/work/xbmc-9999/xbmc/cores/dvdplayer/Codecs/libDVDCSS'
make[1]: *** [libDVDCSS] Error 2

It goes on to build a few more codecs before finally dying.

I have app-text/tetex-3.0_p1-r6 (and app-text/jadetex-3.13-r2) installed.
Comment 42 Xake 2008-10-11 00:58:33 UTC
(In reply to comment #41)
> 
> I have app-text/tetex-3.0_p1-r6 (and app-text/jadetex-3.13-r2) installed.
> 

Try with livetex, it is the recommended default since tetex is a dead project.
that said it seem like I need no tex to compile xbmc...
Comment 43 Peter Fern 2008-10-11 06:42:27 UTC
(In reply to comment #41)
> I had a go at building the xbmc-9999 ebuild, and it dies due to a latex error

I don't have latex installed, please try texlive, and if it still barfs on you, I'll install it and perform some testing... if texlive works for you, I'll add app-text/tetex as a blocker, or patch out the tex stuff from libDVDCSS since it's only documentation...
Comment 44 Peter Fox 2008-10-11 17:33:05 UTC
I tried again with app-text/texlive-2007-r3 installed and it appears to give the same error.
Comment 45 Peter Fox 2008-10-11 18:50:18 UTC
I moved /usr/bin/latex out the way temporarily and it built fine. (Haven't tried using it yet though..)
Comment 46 Peter Fern 2008-10-13 10:32:47 UTC
(In reply to comment #45)
> I moved /usr/bin/latex out the way temporarily and it built fine.

Alright, I'll see if there's a fix upstream at libDVDCSS, otherwise I'll just patch out the tex check, won't get to it for a few days though...
Comment 47 William Johansson 2008-10-13 18:13:03 UTC
I've tried to modify these ebuilds (xbmc-8.10_beta1 and _p15828) and cannot make it to install into /opt instead of all over the system.

Of course, it would be nice with binaries in /usr/bin and so on, but not like the folders /scripts, /web, /userdata and so on...

Why isn't it obeying --prefix=/opt in econf? 
When launching /usr/bin/xbmc it complains about not finding /opt/share/xbmc/xbmc.bin.
Comment 48 Xake 2008-10-13 20:01:31 UTC
(In reply to comment #47)

> Why isn't it obeying --prefix=/opt in econf? 
> When launching /usr/bin/xbmc it complains about not finding
> /opt/share/xbmc/xbmc.bin.


Is not /opt for pre-compiled files?
What does "locate xbmc.bin" give you? Maybe the script /usr/bin/xbmc (which if --prefix=/opt did work would be /opt/bin/xbmc) has a hardcoded and wrong path.
Comment 49 Peter Fern 2008-10-13 23:05:46 UTC
(In reply to comment #47)
> Why isn't it obeying --prefix=/opt in econf? 
> When launching /usr/bin/xbmc it complains about not finding
> /opt/share/xbmc/xbmc.bin.
> 

The XBMC guys' use of autotools isn't complete - their build script does a lot of copying outside the makefile, which is what my ebuilds emulate.  I don't really see a reason to install into /opt, but if you want to, you'll need to modify src_install with some bininto, docinto, etc lines...
Comment 50 Peter Fern 2008-10-20 06:50:44 UTC
Created attachment 169152 [details]
Pegged ebuild

libDVDCSS barfs on all current *tex implementatinos, temporarily disable latex test.
Comment 51 Peter Fern 2008-10-20 06:51:32 UTC
Created attachment 169154 [details]
Trunk version of previous ebuild

libDVDCSS barfs on all current *tex implementatinos, temporarily disable latex test.
Comment 52 thephantom1492 2008-10-22 01:54:29 UTC
Thanks for this ebuild

there is 3 small things I found:
* this package seems to require pcre, not an optional thing
* dev-libs/cyrus-sasl needs USE="ssl" (else can't download via subversion, https...)
* missing dependency: app-i18n/enca
Comment 53 Peter Fern 2008-10-22 04:24:08 UTC
Excuse my slight slight edition of your post

(In reply to comment #52)
> * this package seems to require pcre, not an optional thing
> * missing dependency: app-i18n/enca

I'll verify and add these in the next day or two

> * dev-libs/cyrus-sasl needs USE="ssl" (else can't download via subversion,
> https...)

If this is the case, it should really be handled by the subversion eclass, probably worth submitting a bug for this
Comment 54 Luca Vitucci 2008-10-23 17:16:29 UTC
Thank you for your ebuild, but i found some missing dependencies (I'm compiling XBMC on a new system with only fluxbox and its dependencies)
USE="hal opengl pcre"

-----
checking for main in -lfaac... no
configure: error: == Could not find a required library. Please see README.linux

Missing: media-libs/faac

-----
checking for main in -lenca... no
configure: error: == Could not find a required library. Please see README.linux

Missing: app-i18n/enca

-----
checking for nasm... no
configure: error: == Could not find a required program.  Please see README.linux.

Missing: dev-lang/nasm
Comment 55 Daniel Marmander 2008-10-25 08:27:11 UTC
Beta ebuild and pegged ebuild compiles fine, but XBMCTex does not create the textures as is should. It takes care of .gif, but not .png:

 * Generating textures...
xbmc-logo.png:                            Error creating surface size 0 by 0
unknown-user.png:                         Error creating surface size 0 by 0
thumb-back.png:                           Error creating surface size 0 by 0
sub-suspend-nofocus.png:                  Error creating surface size 0 by 0
sub-suspend-focus.gif:                     128x128  (   28f)->(   15f)  128x128 
 ( 0.00% waste)
sub-shutdown-nofocus.png:                 Error creating surface size 0 by 0

...

If I try to execute XBMCTex manually, it does the same thing.

When compiled from source, without the ebuild, it seems to work fine.

Happends with both 8.10_beta1 and 8.10_p15940.
Comment 56 Peter Fern 2008-10-27 09:08:02 UTC
Sorry, been busy this week, will look into getting the updated ebuild with additional deps, and a fix for the XBMCTex PNG issue either tomorrow or the day after.
Comment 57 Rick Harris 2008-10-28 04:31:45 UTC
Have some problems with the latest xbmc-9999.ebuild

Once the missing dependencies were installed, then everything compiles and the application runs, however video playback segfaults on any video and audio playback segfaults on any MP3, OGG etc.

The xmms patch and accompanying sed'ing of the source to get it to compile (test_libFLAC++, liba52 and libDVDCSS latex problems) are due to running 'eautoreconf' using the wrong version of automake.
The solution is to run aclocal/libtoolize etc. only in the directories it is required and with the correct WANT_AUTOMAKE="<version>".

The video playback segfaulting was due to media-libs/alsa-lib not having been built with 'debug' enabled.
The audio playback segfaulting was due to the eautoreconf problem mentioned above.

Once this was sorted, MP3 playback would not work due to alsa complaining the sound device was busy. This was due to media-libs/libsdl defaulting to use the OSS sound driver if it had been built with the 'oss' USE flag.
The solution is to export SDL_AUDIODRIVER=alsa before running 'xbmc'.

Have updated the ebuild with the following missing dependencies:
        =sys-devel/automake-1.8*
        =sys-devel/automake-1.9*
        =sys-devel/automake-1.10*
        media-libs/faac
        dev-lang/nasm
        app-i18n/enca
        sys-apps/lsb-release

Added automake fix re-write.
Added a check to ensure media-libs/alsa-lib is built_with_use 'debug'.
Added a check to ensure media-libs/libsdl is built_with_use 'alsa'.
Added installation of the xbmc.desktop file and accompanying icon.
/etc/env.d/95xbmc file now gets installed.
Added 'strip-flags' to remove custom CFLAGS and allow upstream to debug.
Added 'export SDL_AUDIODRIVER=alsa' to the installed 'xbmc' wrapper script.
Added 'gnome' USE flag to enable/disable gnome's screensaver in the 'xbmc' wrapper script.

By default the 'xbmc' wrapper script tries to connect via dbus to Gnome's screensaver to inhibit it on run and uninhibit it on exit.
For those that don't run Gnome it's an untidy error message and an unnecessary process.
Have tested fullscreen mode in KDE and xbmc disables the screensaver anyway without the need for dbus calls, so it may not be needed for Gnome either but I'm unable to test for Gnome.

I wasn't able to replicate Daniel Marmander's PNG problem when creating textures.

Ebuild is attached below ...
Comment 58 Rick Harris 2008-10-28 04:33:37 UTC
Created attachment 170055 [details]
xbmc-9999-r1.ebuild
Comment 59 Peter Fern 2008-10-28 07:02:37 UTC
Comment on attachment 169154 [details]
Trunk version of previous ebuild

Thanks Rick!  I still don't have a non-headless Gentoo box as I'm still in the midst of a prolonged house move, so it's been difficult for me to debug, appreciate the effort.

Feel free to update the pegged build if you have the time, since I still believe it to be useful should this ever make it into portage.
Comment 60 Rick Harris 2008-10-29 02:02:04 UTC
No problem, glad I could help :)

Regarding the pegged ebuild, it would be easy enough to update, just take the 9999-r1 ebuild and add the ESVN_REVISION stuff at the top.

But I'm not sure pegged ebuilds are the way to go for if/when this ever gets into portage as official beta sources are being released almost monthly at present (unlike MythTV which uses pegs because of such large periods between releases). I think just having an SVN ebuild and making ebuilds for the betas and RC's would be the way to go.

As we type, upstream are in a feature freeze at the moment just fixing bugs, so another beta or rc1 is on the way very soon.
Comment 61 Per Öberg 2008-11-03 08:30:14 UTC
I tried to install the xbmc ebuilds yesterday but I got some sandbox violations on  "/var/cache/fonts/pk/".

xbmc-8.10_p15940.ebuild installet nicely with minor dependency issues but when i tried moving on to the xbmc-9999-r1.ebuild i get the sandbox violation. 

I have no idea how to debug this but I'm willing to give it a shot with a little help from someone who knows where to begin. 

/var/cache seems harmless, is it possible to exclude it or is that just an ugly hack?
Comment 62 Xake 2008-11-03 09:01:09 UTC
(In reply to comment #61)
> I tried to install the xbmc ebuilds yesterday but I got some sandbox violations
> on  "/var/cache/fonts/pk/".
> 
> xbmc-8.10_p15940.ebuild installet nicely with minor dependency issues but when
> i tried moving on to the xbmc-9999-r1.ebuild i get the sandbox violation. 
> 

This means something has changed in their buildsystem and that thing ignores DESTDIR. What is also intresting with this kind of problems is what Makefile that did this (i.e. what directory make processed while this happened) and if it was during compile or install.
What you can try is to empty your distdir (i.e. "rm -rf /usr/portage/svn-src/xbmc") as this will make xbmc refetch a clean repo.
Comment 63 Per Öberg 2008-11-03 17:52:19 UTC
> What you can try is to empty your distdir (i.e. "rm -rf
> /usr/portage/svn-src/xbmc") as this will make xbmc refetch a clean repo.
> 

I tried this but with the same sandbox issue.


> What is also intresting with this kind of problems is what Makefile
> that did this (i.e. what directory make processed while this happened) and if
> it was during compile or install.

I think it's during the compile. 

Short version of one of the errors: (All of them seems to be related to libDCDCSS/doc)

make[4]: *** [stamp-latex] Error 2
make[4]: Leaving directory `/var/tmp/portage/media-tv/xbmc-9999-r1/work/xbmc-9999/xbmc/cores/dvdplayer/Codecs/libDVDCSS/doc'

I have attatched parts of the build (I cut it because it was HUGE)
Comment 64 Per Öberg 2008-11-03 17:53:16 UTC
Created attachment 170640 [details]
cutout of build.log for sandbox issue
Comment 65 Per Öberg 2008-11-03 21:03:08 UTC
(In reply to comment #64)
> Created an attachment (id=170640) [edit]
> cutout of build.log for sandbox issue
> 

Well, it looks like the problem is that kphatsea generates fonts for the documentation in the wrong place. These are temporary files that you normally don't care about where they go.

I generated the documentation manually by running make in /var/tmp/portage/media-tv/xbmc-9999-r1/work/xbmc-9999/xbmc/cores/dvdplayer/Codecs/libDVDCSS/doc
and then re run the emerge. 

It now works for me but I am unsure of how to solve this permanently without disabling the doc-generation. It seems like there is an environment variable DVIPSFONTS that might be able to change the behavior though.
Comment 66 Rick Harris 2008-11-04 22:27:47 UTC
Probably partially my fault.
I removed Peter Fern's disabling of generating documentation in xbmc/cores/dvdplayer/Codecs/libDVDCSS/ with the thought it was the automake version issue causing this (sorry Peter).

I see now it was due to my system not having Latex installed, so doc generation was skipped over.
Installing latex now to reproduce, would be nice to also have a 'doc' USE flag once this problem is sorted.
Comment 67 Rick Harris 2008-11-05 22:55:59 UTC
Even with latex and doxygen installed I couldn't reproduce the same sandbox violation, everything built fine.
The libDVDCSS docs that it is attempting to create are not really relevant to XBMC anyway, but are an internal library's API docs.
To clarify, the same docs could probably be provided by 'USE="doc" emerge media-libs/libdvdcss'.

For this reason, and because XBMC sources do not ship with any documentation pertaining to the end application, I don't see any need for documentation to be built at all. All the usage documentation for XBMC is on-line and is not distributed with it's sources.

So with that in mind, have updated the ebuild with the following changes:
* Re-added disabling of doc building for libDVDCSS
* Added link to on-line documentation
* Removed dbus screensaver connections for Gnome in 'xbmc' wrapper script and so also 'gnome' USE flag
* Fixed ebuild creating a bunch of empty directories in root directory
* Fixed XBMC's version string to contain the compiled SVN revision
* Prevent the installation of Mac OSX files
* General tidy up

Ebuild attached below ...
Comment 68 Rick Harris 2008-11-05 22:57:33 UTC
Created attachment 170858 [details]
xbmc-9999-r2.ebuild
Comment 69 Cameron Tacklind 2008-11-06 08:36:37 UTC
I just tried emerging xbmc-9999-r2.ebuild and ran into dependancy issues with hal

Configure script died on -lhal check

After enabling hal use it of course got the hal dependency and it is now compiling. It looks like hal is now required?

While writing this, my build failed :-(

On this command:
i686-pc-linux-gnu-g++ -MD -c -O2 -march=i686 -pipe -O2 -DNDEBUG=1 -DDBUS_API_SUBJECT_TO_CHANGE -I/usr/include/dbus-1.0 -I/usr/lib/dbus-1.0/include -I/usr/include/hal -I/usr/include/lzo -O2 -DNDEBUG=1 -DDBUS_API_SUBJECT_TO_CHANGE -I/usr/include/dbus-1.0 -I/usr/lib/dbus-1.0/include -I/usr/include/hal -I/usr/include/lzo -D_LINUX -D_FILE_DEFINED -D_REENTRANT -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64  -DHAS_PCRE -D'SVN_REV="16114"' -DINSTALL_PATH="\"/usr/share/xbmc\"" -DDLL_PATH_LIBCURL="\"/usr/lib/libcurl.so.4\"" -I. -Ilinux -Icores -I../guilib -Iutils -IFileSystem -Ilib/libUPnP/Platinum/ThirdParty/Neptune/Source/Core -Ilib/libUPnP/Platinum/Source/Core -Ilib/libUPnP/Platinum/Source/Devices/MediaServer -Ilib/libUPnP/Platinum/ThirdParty/Neptune/Source/System/Posix -Ilib/libUPnP/Platinum/Source/Devices/MediaConnect -Ilib/libUPnP/Platinum/Source/Devices/MediaRenderer   XBVideoConfig.cpp -o XBVideoConfig.o

Error:
XBVideoConfig.cpp: In member function 'void XBVideoConfig::GetDesktopResolution(int&, int&, float&) const':
XBVideoConfig.cpp:155: error: 'g_xrandr' was not declared in this scope

# emerge --info
Portage 2.2_rc13 (default/linux/x86/2008.0, gcc-4.1.2, glibc-2.8_p20080602-r0, 2.6.27-gentoo-r2 i686)
=================================================================
System uname: Linux-2.6.27-gentoo-r2-i686-Intel-R-_Core-TM-2_CPU_6600_@_2.40GHz-with-glibc2.0
Timestamp of tree: Wed, 05 Nov 2008 23:30:01 +0000
app-shells/bash:     3.2_p39
dev-lang/python:     2.4.4-r13, 2.5.2-r8
dev-python/pycrypto: 2.0.1-r6
dev-util/cmake:      2.6.2
sys-apps/baselayout: 2.0.0
sys-apps/openrc:     0.3.0-r1
sys-apps/sandbox:    1.2.18.1-r3
sys-devel/autoconf:  2.63
sys-devel/automake:  1.8.5-r3, 1.9.6-r2, 1.10.1-r1
sys-devel/binutils:  2.19
sys-devel/gcc-config: 1.4.0-r4
sys-devel/libtool:   2.2.6a
virtual/os-headers:  2.6.27-r2
ACCEPT_KEYWORDS="x86 ~x86"
CBUILD="i686-pc-linux-gnu"
CFLAGS="-O2 -march=i686 -pipe"
CHOST="i686-pc-linux-gnu"
CONFIG_PROTECT="/etc /home/mythtv/"
CONFIG_PROTECT_MASK="/etc/ca-certificates.conf /etc/env.d /etc/fonts/fonts.conf /etc/gconf /etc/gentoo-release /etc/php/apache2-php5/ext-active/ /etc/php/cgi-php5/ext-active/ /etc/php/cli-php5/ext-active/ /etc/revdep-rebuild /etc/terminfo /etc/udev/rules.d"
CXXFLAGS="-O2 -march=i686 -pipe"
DISTDIR="/usr/portage/distfiles"
FEATURES="distlocks parallel-fetch preserve-libs protect-owned sandbox sfperms strict unmerge-orphans userfetch"
GENTOO_MIRRORS="http://gentoo.mirrors.easynews.com/linux/gentoo/ ftp://gentoo.netnitco.net/pub/mirrors/gentoo/source/ ftp://distro.ibiblio.org/pub/linux/distributions/gentoo/ http://distro.ibiblio.org/pub/linux/distributions/gentoo/ http://mirror.csclub.uwaterloo.ca/gentoo-distfiles/"
LDFLAGS="-Wl,-O1"
MAKEOPTS="-j3"
PKGDIR="/usr/portage/packages"
PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --compress --force --whole-file --delete --stats --timeout=180 --exclude=/distfiles --exclude=/local --exclude=/packages"
PORTAGE_TMPDIR="/var/tmp"
PORTDIR="/usr/portage"
PORTDIR_OVERLAY="/usr/local/portage"
SYNC="rsync://rsync.gentoo.org/gentoo-portage"
USE="acl alsa bash-completion berkdb bzip2 cli cracklib crypt cups dri fortran gdbm gpm iconv ipv6 isdnlog midi mudflap mysql ncurses nls nptl nptlonly opengl openmp pam pcre perl php pppd python readline reflection session spl ssl sysfs tcpd unicode x86 xorg xvmc zlib" ALSA_CARDS="ali5451 als4000 atiixp atiixp-modem bt87x ca0106 cmipci emu10k1 emu10k1x ens1370 ens1371 es1938 es1968 fm801 hda-intel intel8x0 intel8x0m maestro3 trident usb-audio via82xx via82xx-modem ymfpci" ALSA_PCM_PLUGINS="adpcm alaw asym copy dmix dshare dsnoop empty extplug file hooks iec958 ioplug ladspa lfloat linear meter mmap_emul mulaw multi null plug rate route share shm softvol" APACHE2_MODULES="actions alias auth_basic authn_alias authn_anon authn_dbm authn_default authn_file authz_dbm authz_default authz_groupfile authz_host authz_owner authz_user autoindex cache dav dav_fs dav_lock deflate dir disk_cache env expires ext_filter file_cache filter headers include info log_config logio mem_cache mime mime_magic negotiation rewrite setenvif speling status unique_id userdir usertrack vhost_alias" ELIBC="glibc" INPUT_DEVICES="keyboard mouse" KERNEL="linux" LCD_DEVICES="bayrad cfontz cfontz633 glk hd44780 lb216 lcdm001 mtxorb ncurses text" USERLAND="GNU" VIDEO_CARDS="nvidia"
Unset:  CPPFLAGS, CTARGET, EMERGE_DEFAULT_OPTS, FFLAGS, INSTALL_MASK, LANG, LC_ALL, LINGUAS, PORTAGE_COMPRESS, PORTAGE_COMPRESS_FLAGS, PORTAGE_RSYNC_EXTRA_OPTS
Comment 70 Rick Harris 2008-11-06 22:09:45 UTC
Looks like x11-libs/libXrandr might also be required now too.
Comment 71 Cameron Tacklind 2008-11-07 00:05:44 UTC
(In reply to comment #70)
> Looks like x11-libs/libXrandr might also be required now too.

I can't tell if I've had libXrandr installed on my first ebuild because I was emerging some more X stuff that could have pulled it. (It was installed when I saw #70 and tried again)

The same error came up on my first try and went away when I turned on the randr use flag. I'm guessing the xbmc dev team isn't staying true to their configure options...?

But now I've got a problem trying to compile guilib/Surface.cpp:

Surface.cpp: In constructor 'Surface::CSurface::CSurface(int, int, bool, Surface::CSurface*, Surface::CSurface*, SDL_Surface*, bool, bool, bool, int)':
Surface.cpp:244: error: 'struct SDL_SysWMinfo' has no member named 'info'

Comment 72 Rick Harris 2008-11-07 00:26:22 UTC
What are your USE flags for media-libs/libsdl ?
Comment 73 Rick Harris 2008-11-07 01:20:51 UTC
Cameron, able to reproduce your latest error when media-libs/libsdl is not built with the 'X' USE flag enabled.
Comment 74 Rick Harris 2008-11-07 12:02:51 UTC
Thanks Cameron, based on your USE flag problems I've done some dependency USE flag auditing with the ebuild updated as follows:

* Added check to ensure media-libs/libsdl is built_with_use 'X'
* Removed 'xrandr' USE flag as is required to build
* Removed 'hal' USE flag as is required to build
* Removed 'pcre' USE flag as is required to build
* Added app-arch/unzip as a dependency
* Removed forcing '+debug' on users as XBMC when crashing writes out the contents of system memory to a coredump file in the user's /home directory usually in the hundreds of MB's, sometimes GB's. I figure we stay consistent to all other packages in portage where if the user has a problem, they are intelligent enough to build with the 'debug' USE flag enabled.
* Added note pointing to bug filing details for upstream (including reminder to enable the 'debug' USE flag).
* Cleaned up redundant inherits

Ebuild attached below ...
Comment 75 Rick Harris 2008-11-07 12:03:58 UTC
Created attachment 170976 [details]
xbmc-9999-r3.ebuild
Comment 76 Per Öberg 2008-11-07 12:59:23 UTC
(In reply to comment #74)

Just wanted to mention the upstreams request for forcing debug in comment #15 to simplify their work.

Comment 77 Daniel Marmander 2008-11-07 18:35:47 UTC
Seems like media-libs/faac and app-i18n/enca should be deps too.

checking for main in -lfaac... no
configure: error: == Could not find a required library. Please see README.linux


checking for main in -lenca... no
configure: error: == Could not find a required library. Please see README.linux

After installing them, it configures and compiles. Still got the same .png issue, though.
Comment 78 Cameron Tacklind 2008-11-07 22:56:34 UTC
Rick, thanks for the new ebuild but now I'm running into problems with libtool. I've been trying to debug this but I'm not really getting anywhere.

..............................
Making all in dvdread
make[5]: Entering directory `/var/tmp/portage/media/xbmc-9999-r3/work/xbmc-9999/xbmc/cores/dvdplayer/Codecs/libdvdnav/src/dvdread'
/bin/sh ../../libtool --tag=CC   --mode=compile gcc -DHAVE_CONFIG_H -I. -I../.. -I../..  -DDVDNAV_COMPILE -I../../src/vm   -O3 -Wall -D_FILE_OFFSET_B
ITS=64 -D_LARGEFILE64_SOURCE -D_XBOX -DHAVE_DVDCSS_DVDCSS_H -MT dvd_reader.lo -MD -MP -MF .deps/dvd_reader.Tpo -c -o dvd_reader.lo dvd_reader.c
../../libtool: line 463: CDPATH: command not found
../../libtool: line 1148: func_opt_split: command not found
libtool: Version mismatch error.  This is libtool 2.2.6, but the
libtool: definition of this LT_INIT comes from an older release.
libtool: You should recreate aclocal.m4 with macros from libtool 2.2.6
libtool: and run autoconf again.
make[5]: *** [dvd_reader.lo] Error 63
..............................

I am also using the new r3

I saw the other comments about the similar issues but no luck...
Comment 79 Peter Fox 2008-11-08 08:41:50 UTC
Is bug 245777 relevant (and bug 245544 ?)
It looks like there are incompatibilities between libtool versions.
I'm still on libtool-1.5.26.
Comment 80 Cameron Tacklind 2008-11-08 11:24:35 UTC
Ok, I remerged libtool which fixed it. I had tried that before but it didn't fix the problem. weird...

Anyway, now I'm getting what looks to be python issues:

................................
Compiling ../Python/Lib/xml/sax/__init__.pyo
Compiling ../Python/Lib/xml/sax/_exceptions.pyo
Compiling ../Python/Lib/xml/sax/expatreader.pyo
Compiling ../Python/Lib/xml/sax/saxutils.pyo
Compiling ../Python/Lib/xml/sax/handler.pyo
Compiling ../Python/Lib/xml/sax/xmlreader.pyo
pushd ../Python/build/lib.linux-i686-2.4/ && \
                zip -0guq ../../../../../../system/python/python24.zip *.so; \
                popd
/var/tmp/portage/media/xbmc-9999-r3/work/xbmc-9999/xbmc/lib/libPython/Python/build/lib.linux-i686-2.4 /var/tmp/portage/media/xbmc-9999-r3/work/xbmc-9999/xbmc/lib/libPython/linux
/var/tmp/portage/media/xbmc-9999-r3/work/xbmc-9999/xbmc/lib/libPython/linux
pushd ../Python/Lib/ && \
                zip -0gurq ../../../../../system/python/python24.zip *.py* bsddb compiler curses distutils email encodings hotshot idlelib lib-old lib-tk logging site-packages xml plat-linux2 -x *.svn* *test*; \
                popd
/var/tmp/portage/media/xbmc-9999-r3/work/xbmc-9999/xbmc/lib/libPython/Python/Lib /var/tmp/portage/media/xbmc-9999-r3/work/xbmc-9999/xbmc/lib/libPython/linux
/var/tmp/portage/media/xbmc-9999-r3/work/xbmc-9999/xbmc/lib/libPython/linux
make[2]: Leaving directory `/var/tmp/portage/media/xbmc-9999-r3/work/xbmc-9999/xbmc/lib/libPython/linux'
make[1]: Leaving directory `/var/tmp/portage/media/xbmc-9999-r3/work/xbmc-9999/xbmc/lib/libPython/linux'
 *
 * ERROR: media/xbmc-9999-r3 failed.
 * Call stack:
 * ebuild.sh, line   49:  Called src_compile
 * environment, line 2384:  Called die
 * The specific snippet of code:
 * emake || die "Make failed!";
 * The die message:
 * Make failed!
 *
 * If you need support, post the topmost build error, and the call stack if relevant.
 * A complete build log is located at '/var/tmp/portage/media/xbmc-9999-r3/temp/build.log'.
 * The ebuild environment file is located at '/var/tmp/portage/media/xbmc-9999-r3/temp/environment'.
 * This ebuild is from an overlay: '/usr/local/portage/'
 *

>>> Failed to emerge media/xbmc-9999-r3, Log file:

>>>  '/var/tmp/portage/media/xbmc-9999-r3/temp/build.log'
.............................

It actually looks like the python compiling finished and something else failed.
I tried to run both pushd/zip/popd commands and they each gave the 2 line outputs correctly. No errors.
Comment 81 Peter Fox 2008-11-09 08:41:38 UTC
The error is not in the snippet you showed. Sometimes the build continues for some time after an error so you have to look back a long way to see the actual error.
Comment 82 Cameron Tacklind 2008-11-10 18:06:49 UTC
Ah, of course.

Building with MAKEOPTS="-j1" made the libtool problem resurface.

I'm still at a loss on how to fix the libtool issue.
Comment 83 Rick Harris 2008-11-12 07:19:45 UTC
OK, I was way off the mark with the WANT_AUTOMAKE version stuff (eats humble
pie).
I see now that the patching and sed'ing is due to upstream updating the source,
but not the autotools files.
So that if say aclocal, autoconf or automake are needed to be run to re-create
aclocal.m4, configure or Makefile.in, the end result can be a bit spotty.
Add a bleeding edge libtool version into the mix and things get trickier.

To cut a long story short, here is an update to the ebuild that should build on
systems using either libtool-1.5* or libtool-2.2*.

Ebuild attached below ...
Comment 84 Rick Harris 2008-11-12 07:20:38 UTC
Created attachment 171513 [details]
xbmc-9999-r4.ebuild
Comment 85 Peter Fern 2008-11-12 09:34:45 UTC
(In reply to comment #83)
> OK, I was way off the mark with the WANT_AUTOMAKE version stuff (eats humble
> pie).
> I see now that the patching and sed'ing is due to upstream updating the source,
> but not the autotools files.

Thanks again for running with this Mike - I still don't have a box I can usefully run this code on for testing.

I know my original ebuild seemed a bit broad-brushed and heavy-handed in it's (excessive) use of eautoreconf, however the reasoning behind it was (aside from the obvious laziness ;) ): since the same problems existed in so many of their libs, and eautoreconf knows when patching is needed - eg when using libtool-?.? - and with this being a live ebuild, and the likelihood of further breakages of this type during development, this (ham-fisted) approach would (semi-)transparently handle any new occurrences without having to update the ebuild.  Indeed, my original live ebuild still compiles cleanly (though is missing some of your tidying, deps, etc) against the current HEAD using -j1 or whatever, with no modification.

Just food for thought, and please excuse the horrible sentence structure, been a long day.
Comment 86 Xake 2008-11-12 09:53:43 UTC
about xbmc-9999-r4.ebuild:

Why does it pull in that many versions of automake? It only needs latest as far as I can see currently (I have only automake-1.5 and 1.10 installed on my system, and eautoreconf worked fine).

Maybe strip-flag only should be run if USE="debug" is specified?
And maybe specify IUSE="+debug" so the default is to merge with debug?
I know the core-dumps are massive, but upstream wants it for some reasons, and you can easily override it by adding -debug to your make.conf 
or your portage.use.

And maybe consider converting to EAPI=2 with the alsa-lib-use-flags and so on?
Comment 87 Daniel Marmander 2008-11-12 19:32:30 UTC
It seems like the png problem has nothing to do with compile, but rather with install. xbms looks as it should if I run it from the work-directory after "ebuild compile", but of some reason, "ebuild install" does not copy all the files.
Comment 88 Cameron Tacklind 2008-11-12 20:32:59 UTC
Sweet.

r4 has fixed everything :-D
Comment 89 Rick Harris 2008-11-13 02:56:41 UTC
(In reply to comment #86)
> about xbmc-9999-r4.ebuild:
> 
> Why does it pull in that many versions of automake? It only needs latest as far
> as I can see currently (I have only automake-1.5 and 1.10 installed on my
> system, and eautoreconf worked fine).

Should be removed, is a remnant of a previous ebuild.

> Maybe strip-flag only should be run if USE="debug" is specified?
> And maybe specify IUSE="+debug" so the default is to merge with debug?
> I know the core-dumps are massive, but upstream wants it for some reasons, and
> you can easily override it by adding -debug to your make.conf 
> or your portage.use.

Excellent idea.

> And maybe consider converting to EAPI=2 with the alsa-lib-use-flags and so on?

EAPI=2 is not supported by stable portage (yet), but can you elaborate on the 'alsa-lib-use-flags' thing and how it relates to EAPI=2 ?
Comment 90 Rick Harris 2008-11-13 02:57:48 UTC
(In reply to comment #87)
> It seems like the png problem has nothing to do with compile, but rather with
> install. xbms looks as it should if I run it from the work-directory after
> "ebuild compile", but of some reason, "ebuild install" does not copy all the
> files.

Are you using the xbmc-9999-r4.ebuild ?
Comment 91 Cameron Tacklind 2008-11-13 03:09:53 UTC
(In reply to comment #90)
> (In reply to comment #87)
> > It seems like the png problem has nothing to do with compile, but rather with
> > install. xbms looks as it should if I run it from the work-directory after
> > "ebuild compile", but of some reason, "ebuild install" does not copy all the
> > files.
> 
> Are you using the xbmc-9999-r4.ebuild ?


I'm guessing this problem manifests as a quite bland looking XBMC interface. I just started mine up and the menu was there. Navigating around a little bit brought me to screens that were trying to get me to do something but the borders and text were nearly all black on black.

(yes, with r4)
Comment 92 Rick Harris 2008-11-13 06:09:19 UTC
Only guessing without more info, but has media-libs/sdl-image been built with the 'gif', 'jpeg' and 'png' USE flags enabled ?
Comment 93 Cameron Tacklind 2008-11-13 10:09:44 UTC
Rebuilding sdl-image with 'gif', 'jpeg', and 'png' (and 'tiff') fixed it! (after a re-emerge of xbmc of course)

Yay! I've been playing around with it now a little bit and I love it! :-D

Maybe a 'autostart' use flag could be implemented similar to MythTV. (Which I had originally installed and realized it wasn't what I wanted...) Hell, if necessary I'd even try to make one :-D
Comment 94 Xake 2008-11-13 10:18:38 UTC
(In reply to comment #89)

> > And maybe consider converting to EAPI=2 with the alsa-lib-use-flags and so on?
> 
> EAPI=2 is not supported by stable portage (yet), but can you elaborate on the
> 'alsa-lib-use-flags' thing and how it relates to EAPI=2 ?
> 

You can do fansy stuff like "media-libs/alsa-lib[debug]" and it requires media-libs/alsa-lib to be built with debug (no more "if built-with-use" and so on). It seems to be a little bit trickier with sdl and joystick since the flags do not have the same name. But for sdl and png/jpg/gif it could save a lot of trouble.:-)


(In reply to comment #93)
> Rebuilding sdl-image with 'gif', 'jpeg', and 'png' (and 'tiff') fixed it!
> (after a re-emerge of xbmc of course)
> 
> Yay! I've been playing around with it now a little bit and I love it! :-D
> 
> Maybe a 'autostart' use flag could be implemented similar to MythTV. (Which I
> had originally installed and realized it wasn't what I wanted...) Hell, if
> necessary I'd even try to make one :-D
> 

I have one working. Copied the pieces from MythTV, removed the wm from .xinitrc, and replaced the line for mythtv with xbmc. Works nice as user xbmc...:-)
Comment 95 Per Öberg 2008-11-13 10:36:22 UTC
(In reply to comment #93 and comment #94)

I hate do discuss this here, there probably is a better place, but Gentoo xbmc users seems to have gathered here.

Now, I see that some of you also use MythTV. I have set up xbmc for MythTV TV watching which works quite nicely. I have, however, problems on getting it to talk to myth about music via the upnp interface. (The mythtv upnp service was working, shortly, when I was trying sharing files from Windows Media player on a friends laptop as well as using my web-radio as a upnp media player, but now I cannot get it to work again)

I compiled using the debug flag but I can't see any debug messages indicating that something is wrong. (Other than the screen saver stuff...)

No one over att xbmc forums seems to recognize the problem. Is it possible that this is a gentoo issue?
Comment 96 Rick Harris 2008-11-13 11:43:56 UTC
(In reply to comment #93)
> Rebuilding sdl-image with 'gif', 'jpeg', and 'png' (and 'tiff') fixed it!
> (after a re-emerge of xbmc of course)
> 
> Yay! I've been playing around with it now a little bit and I love it! :-D

Great! I just came on to post that by compiling sdl-image with those USE flags
disabled, I was able to reproduce your bland GUI problem and also Daniel
Marmander's error ...
xbmc-logo.png:                            Error creating surface size 0 by 0

... but you already have it fixed :)

(In reply to comment #94)
> You can do fansy stuff like "media-libs/alsa-lib[debug]" and it requires
> media-libs/alsa-lib to be built with debug (no more "if built-with-use" and so
> on). It seems to be a little bit trickier with sdl and joystick since the flags
> do not have the same name. But for sdl and png/jpg/gif it could save a lot of
> trouble.:-)

That'd be grand, I can't wait for it to go unkeyworded. I think it must be very
close now.

> (In reply to comment #93)
> > Maybe a 'autostart' use flag could be implemented similar to MythTV. (Which I
> > had originally installed and realized it wasn't what I wanted...) Hell, if
> > necessary I'd even try to make one :-D
>
> I have one working. Copied the pieces from MythTV, removed the wm from
> .xinitrc, and replaced the line for mythtv with xbmc. Works nice as user
> xbmc...:-)

Xake, can you add the 'autostart' USE flag and feature to the
xbmc-9999-r5.ebuild, attach it as xbmc-9999-r6.ebuild and we'll make
xbmc-9999-r5.ebuild obsolete.

Have updated ebuild with the following:
* Added '+debug' to IUSE
* Altered 'strip-flags' to only be used if 'debug' USE is enabled
* Explicitly set EAPI="1" until EAPI="2" in portage goes ~unmasked
* Ensure media-libs/sdl-image is built_with_use {gif,jpeg,png}
* Added a note on howto access a MythTV backend using XBMC
Comment 97 Rick Harris 2008-11-13 11:44:52 UTC
Created attachment 171625 [details]
xbmc-9999-r5.ebuild
Comment 98 Rick Harris 2008-11-13 12:44:03 UTC
(In reply to comment #95)
Works for me.
What exactly is the problem ?
What are the steps you are taking ?

May be worthy to note that for MythTV's UPnP server to show up in 'Add Music Source', you need to click on 'UPnP Devices', surf back by clicking '..', then click on 'UPnP Devices' again.
Comment 99 Daniel Marmander 2008-11-13 16:59:54 UTC
The error message for sdl-image use-flags looks a bit ugly, just cosmetics, but still:

* 
 * ERROR: media-video/xbmc-9999-r5 failed.
 * Call stack:
 *               ebuild.sh, line   49:  Called pkg_setup
 *     xbmc-9999-r5.ebuild, line   78:  Called die
 * The specific snippet of code:
 *                      die "media-libs/sdl-image needs to be built with the \'${flag}\' USE flag enabled."
 *  The die message:
 *   media-libs/sdl-image needs to be built with the \'gif\' USE flag enabled.
Comment 100 Cameron Tacklind 2008-11-13 23:17:14 UTC
(In reply to comment #94)
> I have one working. Copied the pieces from MythTV, removed the wm from
> .xinitrc, and replaced the line for mythtv with xbmc. Works nice as user
> xbmc...:-)

Could you post your xinitrc?
This is the first time I've installed gentoo with a front end and it's nice to finally learn how to setup and use all the X stuff.

Do you not need a WM? (MythTV needs it to play stuff with mplayer right?)
Can you just put "exec xbmc" in xinitrc, similar to how you start KDE (without a login manager)?
Comment 101 Per Öberg 2008-11-13 23:33:18 UTC
(In reply to comment #98)
> Works for me.

Good to know. Up until now I haven't been sure that it was supposed to work =)

> What exactly is the problem ?

I cannot connect to mythbackend via upnp i the music "plugin". It doesn't show up when i try to add upnp source. I had it working shortly as i mentioned in my last post but even though that entry is still in the list of sources i get a "Cannot connect to network server" error when i try to open it.

> What are the steps you are taking ?

I have configured hosts so that my backend has a real name that is recognized. I read somewhere that a faulty dns could be troublesome. In the Video share i have added a myth:// entry which works fine. I seem to remember that I read that they recognize this by upnp so I guess that means that the mythbackend upnp server is alive. I have a upnp capable web-radio that sees mythbackend and that also makes me believe that the mythbackend is alive and working.

I have enabled the upnp client and renderer setting in the network settings tab. I have also tried getting xbmc to output some debug information but I can't find any.

B.t.w where do the debug information go? I don't think I have understood that really. (I get some messages in the prompt when starting but they are related to screensaver and a fault sed command)

 
> 
> May be worthy to note that for MythTV's UPnP server to show up in 'Add Music
> Source', you need to click on 'UPnP Devices', surf back by clicking '..', then
> click on 'UPnP Devices' again.
> 

I just tried this, but no luck. I should mention that I have a laptop with wlan and ethernet and I remember reading something about some upnp client being unable to choose the active one. I use networkmanager to switch between the two and I have tried both. Might the double interfaces be the cause here to?
Comment 102 Per Öberg 2008-11-13 23:55:24 UTC
(In reply to comment #101)
It now worked for a short while  again. I think I'm on to something. But it's weird because my upnp radio is working at all times.

I did two things:
1) Installed 0.21_p18116 (upgrade from 0.21_p17100)
2) Restart the mythbackend

I have been able to reproduce the following a couple of times:

Start xbmc -> connect upnp -> FAIL -> restart backend -> SUCCESS -> restart xbmc -> connect upnp -> FAIL

Restart backend -> start xbmc -> connect upnp -> SUCCESS -> restart xbmc -> FAIL

If I wait to long after backend restart, it fails....


MythBackend log says:

Main::Registering HttpStatus Extension
2008-11-14 00:53:24.437 mythbackend version: 0.21.20080304-1 www.mythtv.org
2008-11-14 00:53:24.438 Enabled verbose msgs: important general
2008-11-14 00:53:24.442 AutoExpire: CalcParams(): Max required Free Space: 3.0 GB w/freq: 15 min
2008-11-14 00:53:26.167 Reschedule requested for id -1.
2008-11-14 00:53:26.841 Scheduled 68 items in 0.7 = 0.02 match + 0.65 place
2008-11-14 00:53:26.850 Seem to be woken up by USER
2008-11-14 00:53:33.209 UPnpMedia: BuildMediaMap - no VideoStartupDir set,  skipping scan.

As i said. It's weird beacuse my upnp radio works...
Comment 103 Rick Harris 2008-11-14 00:01:43 UTC
(In reply to comment #101)
> (In reply to comment #98[edited])
 
> I have enabled the upnp client and renderer setting in the network settings
> tab.

You shouldn't need to do that, leaving them at the default state of 'Off' for each should have audio still work via UPnP from the MythTV backend. At least this is how mine is set.

> B.t.w where do the debug information go? I don't think I have understood that
> really. (I get some messages in the prompt when starting but they are related
> to screensaver and a fault sed command)

You need to enable debug info in XBMC under Settings > System > Enable Debug Logging, then that will give a debug log that can be found in /var/tmp/<user>-xbmc.log

> I should mention that I have a laptop with wlan
> and ethernet and I remember reading something about some upnp client being
> unable to choose the active one. I use networkmanager to switch between the two
> and I have tried both. Might the double interfaces be the cause here to?

Not really my area of knowledge, but I guess it could be possible if both interfaces use entirely different subnets, are 'up', and the UPnP client chooses the wrong one. Could try and troubleshoot that by manually bringing down the interface that it shouldn't be communicating on via 'ifconfig'.
Comment 104 Rick Harris 2008-11-15 10:45:52 UTC
Created attachment 171788 [details]
xbmc-plugininstaller-9999.ebuild

Here is an ebuild for Google's XBMC plugin repo. installer
Most plugins work, a lot still don't, but errors seem mostly due to the nature of scrapers and the ever changing layouts of the sites they are attempting to scrap (if they exist at all).

Requires the XBMC running user to symlink by
'ln -s /usr/share/xbmc/plugins/programs ~/.xbmc/plugins/programs'

Have at it :D
Comment 105 Rick Harris 2008-11-16 11:32:31 UTC
Created attachment 171958 [details]
xbmc-8.10.ebuild

XBMC 8.10 stable was released a couple of days ago.
Have updated the xbmc-9999-r5 ebuild to xbmc-8.10 with the following addtions:
* Added 'autostart' feature, modified from MythTV (minus mingetty and an xinitrc)
* Tidied error messages (thanks Daniel)
* Fixed typo in note on howto connect to a MythTV backend (should be myth:// not mythtv://)
Comment 106 Xake 2008-11-16 15:36:34 UTC
(In reply to comment #105)
> * Added 'autostart' feature, modified from MythTV (minus mingetty and an
> xinitrc)

I should take a look into what you have done, but one big thing hit me directly:
"gcc -w -o autologinxbmc autologinxbmc.c" during src_install?
How ever you want to do it, that should be in src_compile.
And this is so wrong. You just broke crosscompile and a lot of other portage features for this file. At least use $(CC)?

What is the advantage of using this solution more then you do not need mingetty?
And why remove the xinitrc?

(In reply to comment #100)
> Could you post your xinitrc?

If I find the time later in the week I may do it.

> 
> Do you not need a WM? (MythTV needs it to play stuff with mplayer right?)
> Can you just put "exec xbmc" in xinitrc, similar to how you start KDE (without
> a login manager)?
> 

AFAIK you do not need it when you only run XBMC. MythTV runs external apps for some files and so one (actually I did run MythTV without a windowmanager too since I had problems with the wm), while xbmc runs everything in one window.
Comment 107 Rick Harris 2008-11-16 22:48:30 UTC
(In reply to comment #106)
> (In reply to comment #105)

> I should take a look into what you have done,

Thanks, would be much appreciated if you could test it.

> but one big thing hit me directly:
> "gcc -w -o autologinxbmc autologinxbmc.c" during src_install?
> How ever you want to do it, that should be in src_compile.
> And this is so wrong. You just broke crosscompile and a lot of other portage
> features for this file. At least use $(CC)?

Very good point. I assume you really mean $(tc-getCC), this is an easy fix.

> What is the advantage of using this solution more then you do not need
> mingetty?
Why install yet another getty when we already have a perfectly functional system getty in 'agetty' ?

> And why remove the xinitrc?
Installing our own xinitrc is not necessary as we run xbmc directly from xinit (rather than from within the xinitrc).
The flipside of this is that it probably provides an easier path for users to customise the X session, so could re-instate.

Once again Xake, do feel free to update the ebuild as you see fit :)
Comment 108 Xake 2008-11-17 08:45:19 UTC
(In reply to comment #107)
> Very good point. I assume you really mean $(tc-getCC), this is an easy fix.

Yes, that was what I meant ( think). Have played too much with dumb Makefile...:-)

> 
> > What is the advantage of using this solution more then you do not need
> > mingetty?
> Why install yet another getty when we already have a perfectly functional
> system getty in 'agetty' ?
> 

I guessed your goal was something like that. And yes it can be a good goal.

> > And why remove the xinitrc?
> Installing our own xinitrc is not necessary as we run xbmc directly from xinit
> (rather than from within the xinitrc).
> The flipside of this is that it probably provides an easier path for users to
> customise the X session, so could re-instate.

I think I will update the 9999.ebuild into a EAPI=2 when I find the time for it and I think I will go for your autologin, but reinstate the xinitrc.
Comment 109 Mathy Vanvoorden 2008-11-17 23:17:59 UTC
The 8.10 ebuild compiles and runs on my amd64 box. It is still compiling on my i686 box and I hope that this time it won't die on the libtool stuff :-)
Comment 110 Mathy Vanvoorden 2008-11-21 22:57:28 UTC
Xbmc now also compiles and runs fine on my i686 box. I also ran into the white screen bug with older nvidia card that is discussed on many sites and I solved it by using the workaround (bind x to next instead of stop) also described on those sites.

Thanks for a great ebuild!
Comment 111 Balazs Nemeth 2008-11-25 09:10:23 UTC
(In reply to comment #109)
> The 8.10 ebuild compiles and runs on my amd64 box. It is still compiling on my
> i686 box and I hope that this time it won't die on the libtool stuff :-)
> 

It compiles fine on my i686 and amd64 box. Thank you!
Comment 112 Daniel Plaza 2008-11-25 12:08:00 UTC
Created attachment 173332 [details]
Updated ebuild for xbmc-8.10

  Updated ebuild for xbmc 8.10. Contains minor cosmetic fixes.
Comment 113 Craig Andrews gentoo-dev 2008-12-03 03:34:57 UTC
XBMC includes an internal copy of Python. I know that flameeyes is not going to like that, and I don't know what the official gentoo policy for ebuilds in the official tree are on such internal copies. I just wanted to point it. http://xbmc.org/trac/ticket/5416
Comment 114 Craig Andrews gentoo-dev 2008-12-03 04:10:27 UTC
Also, after emerging this ebuild, portage reported a *lot* of QA issues. I've created a bug with XBMC for that: http://xbmc.org/trac/ticket/5417
Comment 115 Xake 2008-12-03 11:58:34 UTC
(In reply to comment #114)
> Also, after emerging this ebuild, portage reported a *lot* of QA issues. I've
> created a bug with XBMC for that: http://xbmc.org/trac/ticket/5417
> 

Alot of those problems comes from the handling of external libraries.
XBMC comes with *a lot* of external libraries in its tree, some a little changed, some not changed at all and some has changes only needed for xbox. And many of them are some years behind, and some of them are strangely dealt with to conform with xbmc's vfs. I know that if you like to you may send them patches to use system libraries instead as long as it does not brek their vfs.
I encourage this, as that would greatly help with the complexity of this build and also make it somewhat faster to compile as most people already has libraries as ffmpeg, libdvdcss, libfaac, libmad and so on on their systems already. It eould also be easier to debug as if a file works in for example totem (which uses system ffmpeg) and not in xbmc you currently has no idea if it is xbmc's ffmpeg that is broken...
Comment 116 Berend Dekens 2008-12-03 16:52:26 UTC
Works for me on amd64 (Core 2 Duo) like a charm!

On the remark to use system libs - I would make it a use flag: on my HTPC I only installed XBMC which has all needed libs with all required options build in. If I need to install some 30 ebuilds with all sorts of use flags or select to just use everything XBMC has, I would use the latter...
Comment 117 Craig Andrews gentoo-dev 2008-12-03 22:43:27 UTC
XBMC stores executables in /tmp, so if you have secured your /tmp as noexec, you won't be able to do something (like run plugins). http://xbmc.org/trac/ticket/5428
Comment 118 Xake 2008-12-04 18:25:35 UTC
(In reply to comment #116)
> Works for me on amd64 (Core 2 Duo) like a charm!
> 
> On the remark to use system libs - I would make it a use flag: on my HTPC I
> only installed XBMC which has all needed libs with all required options build
> in. If I need to install some 30 ebuilds with all sorts of use flags or select
> to just use everything XBMC has, I would use the latter...
> 

Oh, you sound just like you should run ubuntu+xbmc.;)

A use-flag for external deps can be intresting. However using externals has its advantages. Take a look at xbmc's ffmpeg, and you will find out that it is pretty old, with the holes and with old/missing support for formats. And while there are some of them, a lot of the packages does not have truckloads of use-flags (except python and ffmpeg that is. And maybe curl). And when considering things like security holes in python/libcurl/ffmpeg I think I should prefer a external one.
Comment 119 Rick Harris 2008-12-05 10:52:57 UTC
(In reply to comment #118)
> A use-flag for external deps can be intresting. However using externals has its
> advantages. Take a look at xbmc's ffmpeg, and you will find out that it is
> pretty old, with the holes and with old/missing support for formats. And while
> there are some of them, a lot of the packages does not have truckloads of
> use-flags (except python and ffmpeg that is. And maybe curl). And when
> considering things like security holes in python/libcurl/ffmpeg I think I
> should prefer a external one.

Using existing system libs is definitely a good thing.
It's something that is suited to Gentoo's source based installation and is already being considered upstream, however it's hard to do for ffmpeg.

With regards to ffmpeg, 'new' very often does not mean better.
FFmpeg does not have a release cycle and is simply an SVN tree that breaks often, so the holes and support for different formats often drops in and out for weeks or months at a time.
Which is why most open source projects that use ffmpeg distribute their own ffmpeg snapshot (see MPlayer and others), easier to develop on and support that way.

That said, it's most certainly do-able to use system libs for the ../Codecs/.. directories to use existing system libs, at least as a start.
Comment 120 Peter Fern 2008-12-05 11:26:17 UTC
External lib support is really an upstream responsibility, and on their roadmap I believe... 
Comment 121 Rick Harris 2008-12-05 20:58:19 UTC
Created attachment 174353 [details]
xbmc-9999.ebuild

Re-instate SVN ebuild.
Comment 122 Craig Andrews gentoo-dev 2008-12-07 02:48:44 UTC
The XBMC project is debating whether to use external dependencies, or always statically compile dependencies from their own source tree.

I suggest anyone with an opinion (and I bet almost all Gentoo users/dev's do!) comment in their forums: http://xbmc.org/forum/showthread.php?p=252592
Comment 123 Anton Bolshakov 2008-12-08 22:51:16 UTC
it seems that the upstream has released proper tar balls:
http://sourceforge.net/project/showfiles.php?group_id=87054&package_id=283294
Please adjust the 8.10 ebuild.
Comment 124 Craig Andrews gentoo-dev 2008-12-10 02:47:58 UTC
(In reply to comment #119)
> That said, it's most certainly do-able to use system libs for the ../Codecs/..
> directories to use existing system libs, at least as a start.
I would love to see that happen. If someone can give me some hints (I am a software engineer... I'm just overwhelmed by the complexity of this project) as to how to de-bundle one of these libraries, I'd really appreciate it!
Comment 125 Peter Fern 2008-12-10 04:43:24 UTC
It's worth noting that the libraries may be modified (and some of them definitely are, like python), so this may not be an option without serious work.
Comment 126 Xake 2008-12-10 08:53:51 UTC
(In reply to comment #124)
> (In reply to comment #119)
> > That said, it's most certainly do-able to use system libs for the ../Codecs/..
> > directories to use existing system libs, at least as a start.
> I would love to see that happen. If someone can give me some hints (I am a
> software engineer... I'm just overwhelmed by the complexity of this project) as
> to how to de-bundle one of these libraries, I'd really appreciate it!
> 

Take a look at XBMC/xbmc/DllPaths_generated.h.in, exchange a lib, see what breaks, fix it.

If you do "grep XBOX * -R" you will find most changes in the sourcecode.

(In reply to comment #125)
> It's worth noting that the libraries may be modified (and some of them
> definitely are, like python), so this may not be an option without serious
> work.
> 

Well, I do not htink python is THAT much changes, but xbmc needs to have its handling of python changed to get it to find system-installed modules.
On that must be handled with care tho is ffmpeg, as it has never had a stable api/abi....
Comment 127 Rick Harris 2008-12-12 03:08:16 UTC
Created attachment 175028 [details]
xbmc-8.10-r2.ebuild

Adjusted 8.10 ebuild to use source tarballs instead of SVN branch (thanks Anton)
All the usual autotool's files + libtool hacks still apply with one more interesting quirk:
All the configure and perl scripts lack being executable (+x) in the source tarball.
Comment 128 Rick Harris 2008-12-12 08:01:58 UTC
Created attachment 175034 [details]
xbmc-8.10-r2.ebuild

Fix my silly mistake doing:
find . -name configure.ac -exec echo 'AC_PROG_CXX' >> {} \;
Comment 129 Rick Harris 2008-12-15 10:57:26 UTC
Created attachment 175317 [details]
xbmc-8.10-r3.ebuild

Re-instate the use of 'strip-flags' regardless of whether 'USE=debug' is set or not.
This solves the mysterious segfaulting on MP3 playback when USE="-debug" is set.

Xake, it seemed like a good idea at the time (see comments #86 + #96).

For reference, the segfault occurs in MACDll-i486-linux.so when it is loaded whenever an MP3 file is opened.
Comment 130 Rick Harris 2008-12-25 09:22:07 UTC
Created attachment 176341 [details]
xbmc-8.10-r4.ebuild

Add patch from http://xbmc.org/trac/attachment/ticket/5296/ to allow reading files from data DVD discs.
Comment 131 Rick Harris 2008-12-25 09:23:24 UTC
Created attachment 176344 [details]
xbmc.readsector.patch

Patch from http://xbmc.org/trac/attachment/ticket/5296/ to allow reading files from data DVD dics.
Comment 132 Rick Harris 2008-12-25 09:28:24 UTC
Created attachment 176346 [details]
xbmc-9999-r1.ebuild

Update to include 'strip-flags' fix and data DVD disc support.
Comment 133 Rick Harris 2008-12-27 20:59:59 UTC
Created attachment 176540 [details]
xbmc-plugininstaller-9999-r1.ebuild
Comment 134 alex 2008-12-31 03:45:00 UTC
Compilation fails with no apparent reason:
--
In file included from mputils.h:4,
                 from mputils.c:3:
mp_msg.h:14:7: warning: no newline at end of file
 gcc -DHAVE_CONFIG_H -I. -I.. -fPIC -I/usr/include/freetype2 -MT mputils.lo -MD -MP -MF .d                                                                   eps/mputils.Tpo -c mputils.c -o mputils.o >/dev/null 2>&1
mv -f .deps/mputils.Tpo .deps/mputils.Plo
/bin/sh ../libtool --tag=CC   --mode=link gcc  -fPIC   -I/usr/include/freetype2   -version                                                                   -info 1:0:0 -Wl,-O1 -o libass.la -rpath /usr/local/lib ass.lo ass_cache.lo ass_font.lo ass                                                                   _fontconfig.lo ass_render.lo ass_utils.lo ass_bitmap.lo ass_library.lo mputils.lo  -lenca                                                                    -lfontconfig   -lfreetype -lz
gcc -shared  .libs/ass.o .libs/ass_cache.o .libs/ass_font.o .libs/ass_fontconfig.o .libs/a                                                                   ss_render.o .libs/ass_utils.o .libs/ass_bitmap.o .libs/ass_library.o .libs/mputils.o  /usr                                                                   /lib64/libenca.so -L/usr/lib /usr/lib64/libfontconfig.so -L/usr/lib64 /usr/lib64/libfreety                                                                   pe.so -lz  -Wl,-O1 -Wl,-soname -Wl,libass.so.1 -o .libs/libass.so.1.0.0
(cd .libs && rm -f libass.so.1 && ln -s libass.so.1.0.0 libass.so.1)
(cd .libs && rm -f libass.so && ln -s libass.so.1.0.0 libass.so)
ar cru .libs/libass.a  ass.o ass_cache.o ass_font.o ass_fontconfig.o ass_render.o ass_util                                                                   s.o ass_bitmap.o ass_library.o mputils.o
ranlib .libs/libass.a
creating libass.la
(cd .libs && rm -f libass.la && ln -s ../libass.la libass.la)
make[4]: Leaving directory `/var/tmp/portage/media-video/xbmc-8.10-r4/work/XBMC/xbmc/lib/l                                                                   ibass/libass'
make[4]: Entering directory `/var/tmp/portage/media-video/xbmc-8.10-r4/work/XBMC/xbmc/lib/                                                                   libass'
make[4]: Leaving directory `/var/tmp/portage/media-video/xbmc-8.10-r4/work/XBMC/xbmc/lib/l                                                                   ibass'
make[3]: Leaving directory `/var/tmp/portage/media-video/xbmc-8.10-r4/work/XBMC/xbmc/lib/l                                                                   ibass'
make[2]: Leaving directory `/var/tmp/portage/media-video/xbmc-8.10-r4/work/XBMC/xbmc/lib/l                                                                   ibass'
cp ../libass/.libs/libass.so ../../../../system/players/dvdplayer/libass-x86_64-linux.so
make[1]: Leaving directory `/var/tmp/portage/media-video/xbmc-8.10-r4/work/XBMC/xbmc/lib/l                                                                   ibass/xbmc'
 *
 * ERROR: media-video/xbmc-8.10-r4 failed.
 * Call stack:
 *               ebuild.sh, line   49:  Called src_compile
 *             environment, line 2945:  Called die
 * The specific snippet of code:
 *       emake || die "Make failed!";
 *  The die message:
 *   Make failed!
--
Should I report this upstream?
Comment 135 Rick Harris 2008-12-31 05:21:05 UTC
(In reply to comment #134)
> Compilation fails with no apparent reason:

The error was further up in the output, you'll have to look hard for it and it could be several hundred lines up.
One way to ease this is to compile with MAKEOPTS="-j1" so that the output of several compile jobs does not get inter-mingled.

Hard to know whether it's an upstream, Gentoo or user's system problem without seeing the real error.
Comment 136 alex 2008-12-31 08:39:39 UTC
> The error was further up in the output
Ok, here it is:
--
/var/tmp/portage/media-video/xbmc-8.10-r4/work/XBMC/xbmc/cores/paplayer/paplayer_linux.cpp:357: undefined reference to `snd_strerror'
/var/tmp/portage/media-video/xbmc-8.10-r4/work/XBMC/xbmc/cores/paplayer/paplayer_linux.cpp:368: undefined reference to `snd_strerror'
xbmc/cores/paplayer/paplayer_linux.o:/var/tmp/portage/media-video/xbmc-8.10-r4/work/XBMC/xbmc/cores/paplayer/paplayer_linux.cpp:371: more undefined references to `snd_strerror' follow
xbmc/cores/paplayer/paplayer_linux.o: In function `PAPlayer::ProcessPAP()':
/var/tmp/portage/media-video/xbmc-8.10-r4/work/XBMC/xbmc/cores/paplayer/paplayer_linux.cpp:630: undefined reference to `snd_pcm_pause'
/var/tmp/portage/media-video/xbmc-8.10-r4/work/XBMC/xbmc/cores/paplayer/paplayer_linux.cpp:587: undefined reference to `snd_pcm_pause'
xbmc/cores/paplayer/paplayer_linux.o: In function `PAPlayer::OpenFile(CFileItem const&, CPlayerOptions const&)':
/var/tmp/portage/media-video/xbmc-8.10-r4/work/XBMC/xbmc/cores/paplayer/paplayer_linux.cpp:189: undefined reference to `snd_pcm_reset'
/var/tmp/portage/media-video/xbmc-8.10-r4/work/XBMC/xbmc/cores/paplayer/paplayer_linux.cpp:192: undefined reference to `snd_pcm_pause'
xbmc/cores/paplayer/paplayer_linux.o: In function `PAPlayer::WaitForStream()':
/var/tmp/portage/media-video/xbmc-8.10-r4/work/XBMC/xbmc/cores/paplayer/paplayer_linux.cpp:1129: undefined reference to `snd_pcm_wait'
collect2: ld returned 1 exit status
make: *** [xbmc.bin] Error 1
--

alsa-related or smth. Don't you know whether there are any requirements on ALSA_PCM_PLUGINS for alsa-lib?
Comment 137 Xake 2008-12-31 08:59:18 UTC
(In reply to comment #136)
> > The error was further up in the output
> Ok, here it is:
> --

Post the build.log instead, as you do not have the actual comment that breaks here. But first: have you tried compile xbmc with MAKEOPTS="-j1"? My system can have problems with that sometimes.
Comment 138 alex 2008-12-31 14:02:18 UTC
Created attachment 176956 [details]
log of failing install
Comment 139 alex 2008-12-31 14:04:24 UTC
Yes, I'm compiling with -j1 for now.
Comment 140 Cameron Tacklind 2008-12-31 23:00:18 UTC
It would seem that this bug has strayed off topic.

We seem to have a basically working ebuild. What does it take to get this into the portage tree? That is what I would assume the final goal was for this request in the first place.

For all the current small issues, maybe a new bug should be filled?
Comment 141 Rick Harris 2009-01-02 09:12:25 UTC
(In reply to comment #138)
> Created an attachment (id=176956) [edit]
> log of failing install
> 

What version of media-libs/alsa-lib do you have installed and with what USE flags ?
Comment 142 Rick Harris 2009-01-02 09:16:37 UTC
(In reply to comment #140)
> It would seem that this bug has strayed off topic.
> 
> We seem to have a basically working ebuild. What does it take to get this into
> the portage tree? That is what I would assume the final goal was for this
> request in the first place.
> 
> For all the current small issues, maybe a new bug should be filled?
> 

I'm all for seperate bugs as it keeps things easier to read and sorted into relevance, but at the same time I don't think it would be proper to pollute Gentoo's bugzilla with bugs on ebuilds that are not in the portage tree.
Comment 143 Rick Harris 2009-01-02 09:19:18 UTC
With regard to getting the ebuild into the portage tree, that would require an existing dev. take up the challenge of maintaining it or for someone wishing hard enough for it to be in the main tree to volunteer their time and service to becoming a Gentoo dev.
Comment 144 alex 2009-01-02 11:23:50 UTC
> What version of media-libs/alsa-lib do you have installed and with what USE
> flags ?
I've tried both 1.0.16 & 1.0.18. Now I'm on .18. Turns out, it was a matter of the right 'ALSA_PCM_PLUGINS' combination. For now I have all of them enabled, which finally made XBMC compile. It could be that 'linear' is the key one as it also prevented alsa-tools from compiling. 
Comment 145 Marios Andreopoulos 2009-01-03 10:24:20 UTC
(In reply to comment #130)
> Created an attachment (id=176341) [edit]
> xbmc-8.10-r4.ebuild
> 
> Add patch from http://xbmc.org/trac/attachment/ticket/5296/ to allow reading
> files from data DVD discs.
> 

Hello, thanks for this nice ebuild!

I came upon a small problem. The ebuild doesn't work with the most current (testing) version of fribidi(-0.19.1), though it works with the previous testing version of fribidi(-0.10.9).
The problem happens during the ./configure process. Propably it is the same problem mentioned (and solved) in Fedora mailing lists: http://linux.derkeiler.com/Mailing-Lists/Fedora/2008-09/msg03065.html
Comment 146 Craig Andrews gentoo-dev 2009-01-04 06:36:34 UTC
(In reply to comment #143)
> With regard to getting the ebuild into the portage tree, that would require an
> existing dev. take up the challenge of maintaining it or for someone wishing
> hard enough for it to be in the main tree to volunteer their time and service
> to becoming a Gentoo dev.
> 

I don't think XBMC, in its current state, will be accepted into portage because it violates Gentoo's static libraries policy. Diego (flameeyes) has been writing about this issue for a long time, as the static issue plagued xine-lib for quite a while. Check out his most recent blog post for more: http://blog.flameeyes.eu/2009/01/02/bundling-libraries-for-despair-and-insecurity
Comment 147 alex 2009-01-04 16:50:02 UTC
Rick, would you care to create an overlay for this ebuild?
Comment 148 Rick Harris 2009-01-04 23:52:14 UTC
(In reply to comment #145)
> I came upon a small problem. The ebuild doesn't work with the most current
> (testing) version of fribidi(-0.19.1), though it works with the previous
> testing version of fribidi(-0.10.9).
> The problem happens during the ./configure process. Propably it is the same
> problem mentioned (and solved) in Fedora mailing lists:
> http://linux.derkeiler.com/Mailing-Lists/Fedora/2008-09/msg03065.html

The problem we have is different to that posted on the Fedora mailing list.

For Fedora it is a packaging problem where they have installed the fribidi include *.h files in ${prefix}/lib/libfribidi/ instead of ${prefix}/include/fribidi/ so they now need to patch every bit of source that references those include files.

For us the first problem is that in the 0.19 fribidi series, the include files have been renamed by changing underscores to dashes (eg. fribidi_char_sets.h gets changed to fribidi-char-sets.h).
The second problem is that the 0.19 and 0.10 series are not API/ABI compatible even though http://fribidi.org/ says they 'should' be, they also concede that 0.19 still requires testing before being considered a replacement for the 0.10 series.
If XBMC sources are made to point to the new include filenames, then it will fail at compile time anyway with this error due to the API/ABI incompatibilities -> http://xbmc.org/forum/showthread.php?t=40866

The solution is to make xbmc depend only on versions of fribidi smaller than 0.19*.
Upstream have indicated that they intend to ditch fribidi all together anyway in favour of another lib that provides the same function plus more.

Revised ebuilds attached below ...
Comment 149 Rick Harris 2009-01-04 23:53:16 UTC
Created attachment 177426 [details]
xbmc-8.10-r5.ebuild
Comment 150 Rick Harris 2009-01-04 23:53:50 UTC
Created attachment 177427 [details]
xbmc-9999-r2.ebuild
Comment 151 Rick Harris 2009-01-05 01:26:25 UTC
(In reply to comment #146)
> I don't think XBMC, in its current state, will be accepted into portage because
> it violates Gentoo's static libraries policy. Diego (flameeyes) has been
> writing about this issue for a long time, as the static issue plagued xine-lib
> for quite a while. Check out his most recent blog post for more:
> http://blog.flameeyes.eu/2009/01/02/bundling-libraries-for-despair-and-insecurity

Diego makes valid points and has set some important precedents in getting xine-lib's patches and others pushed into the current FFmpeg SVN tree.

It is XBMC's goal eventually to use system libs where as much as possible.
I don't speak for upstream, but it might never totally use the system libs as they may have been changed either too much in XBMC's sources, have been changed in such a way as to never be accepted by their respective upstream developers, or that XBMC integrally relies on a distributed source's outdated API.

Some XBMC distributed sources that I've stumbled across that have had unique changes are python, ffmpeg and libdvdread, I'm almost certain there must be more.

We could do a major patch up of the XBMC sources to use system libs, but then we also run the danger of not being supported by upstream when problems arise as this could be seen as an unsupported fork.
It is also against Gentoo's policy to run with patches that have not come from upstream, adding the data DVD discs patch has already violated this.

So based on current Gentoo policy and XBMC sources, this ebuild may never make it into portage.

(In reply to comment #147)
> Rick, would you care to create an overlay for this ebuild?

No. I enjoy the current medium of bugs.gentoo.org where anyone with an interest, bugfix or idea can update the ebuild themselves without the need for special privileges, note that the original ebuild is not mine.

It also affords me the ability to bail out whenever the need arises (my 2nd child is due any day now) :D
Comment 152 Andrew Saunders 2009-01-19 02:00:58 UTC
I was having this problem compiling :
In file included from GUIButtonControl.cpp:26:
../xbmc/utils/CharsetConverter.h:31:39: error: fribidi/fribidi_char_sets.h: No such file or directory 

listed on this page is the fix :
http://www.howtoforge.com/installing-xbmc-on-fedora-9-i386

which is :
"cd xbmc ; ln -s lib/libfribidi/ fribidi"

before compiling. 
Comment 153 Xake 2009-01-19 12:02:36 UTC
(In reply to comment #152)
> I was having this problem compiling :
> In file included from GUIButtonControl.cpp:26:
> ../xbmc/utils/CharsetConverter.h:31:39: error: fribidi/fribidi_char_sets.h: No
> such file or directory 
> 
> listed on this page is the fix :
> http://www.howtoforge.com/installing-xbmc-on-fedora-9-i386
> 
> which is :
> "cd xbmc ; ln -s lib/libfribidi/ fribidi"
> 
> before compiling. 
> 

...or depending on libfribidi-1.10 until xbmc is fixed upstream.
Comment 154 Xake 2009-01-19 14:24:21 UTC
Tell me agains why we do have USE="ccache" as it enables nothing that FEATURES="ccache" does not.
Also FEATURES="-ccache" USE="ccache" emerge xbmc still does not use ccache since portage sets CCACHE_DISABLE=1 to force packets to not use ccache for concistency.
The only thing USE="ccache" may lead to currently is that the compilation can take longer time since ccache may be invoked twice.
Comment 155 morgan wahl 2009-01-28 05:23:41 UTC
I've just compiled this on my amd64 system. It compiled and installed fine, but with this warning:
 * 
 * QA Notice: Package has poor programming practices which may compile
 *            but will almost certainly crash on 64bit architectures.
 * 
 * Function `completion_matches' implicitly converted to pointer at /var/tmp/portage/media-video/xbmc-8.10-r5/work/XBMC/xbmc/lib/libPython/Python/Modules/readline.c:683
 * 
 *  Please file a bug about this at http://bugs.gentoo.org/
 *  with the maintaining herd of the package.
 * 

When i try to run `xbmc`, the screen changes resolutions briefly, then i get a segmentation fault.

running `xbmc --help` gives:
Usage: /usr/share/xbmc/xbmc.bin [OPTION]... [FILE]...

Arguments:
  -fs			Runs XBMC in full screen
  --standalone		XBMC runs in a stand alone environment without a window 
			manager and supporting applications. For example, that
			enables network settings.
  --legacy-res		Enables screen resolutions such as PAL, NTSC, etc.
CRITSEC[0x11b9f28]: Trying to enter destroyed section.
CRITSEC[0x11b9f28]: Trying to leave destroyed section.

should i post more info?
Comment 156 Xake 2009-01-28 08:38:09 UTC
Maybe time to have a (In reply to comment #155)

> should i post more info?
> 
Go to http://forums.gentoo.org/viewtopic-p-5427989.html

Please follow the instructions under troubleshooting and post what you find out there.
Comment 157 morgan wahl 2009-01-28 14:29:42 UTC
(In reply to comment #156)
> Maybe time to have a (In reply to comment #155)
> 
> > should i post more info?
> > 
> Go to http://forums.gentoo.org/viewtopic-p-5427989.html
> 
> Please follow the instructions under troubleshooting and post what you find out
> there.
> 

thanks for the help. I was able to open the core dump in gdb and noticed it crashed in a function having to do with GL. turns out, i didn't even have glx loaded into my X-server. runs fine with 'glx' loaded. i guess i don't use opengl that often...
Comment 158 DaggyStyle 2009-02-12 07:23:00 UTC
(In reply to comment #155)
> I've just compiled this on my amd64 system. It compiled and installed fine, but
> with this warning:
>  * 
>  * QA Notice: Package has poor programming practices which may compile
>  *            but will almost certainly crash on 64bit architectures.
>  * 
>  * Function `completion_matches' implicitly converted to pointer at
> /var/tmp/portage/media-video/xbmc-8.10-r5/work/XBMC/xbmc/lib/libPython/Python/Modules/readline.c:683
>  * 
>  *  Please file a bug about this at http://bugs.gentoo.org/
>  *  with the maintaining herd of the package.
>  * 
> 
> When i try to run `xbmc`, the screen changes resolutions briefly, then i get a
> segmentation fault.
> 
> running `xbmc --help` gives:
> Usage: /usr/share/xbmc/xbmc.bin [OPTION]... [FILE]...
> 
> Arguments:
>   -fs                   Runs XBMC in full screen
>   --standalone          XBMC runs in a stand alone environment without a window 
>                         manager and supporting applications. For example, that
>                         enables network settings.
>   --legacy-res          Enables screen resolutions such as PAL, NTSC, etc.
> CRITSEC[0x11b9f28]: Trying to enter destroyed section.
> CRITSEC[0x11b9f28]: Trying to leave destroyed section.
> 
> should i post more info?
> 

same here, how to fix this?
Comment 159 Xake 2009-02-12 08:36:07 UTC
(In reply to comment #158)
Please read the replies he got. 
And please do not fill this bug about the XBMC _ebuild_ with runtime problems, that is what we got the forum for.
Comment 160 DaggyStyle 2009-02-12 13:59:27 UTC
(In reply to comment #159)
> (In reply to comment #158)
> Please read the replies he got. 
> And please do not fill this bug about the XBMC _ebuild_ with runtime problems,
> that is what we got the forum for.
> 

I didn't got the chance to run it, I thought it didn't even installed.
it runs ok.

Comment 161 Andrew Saunders 2009-02-19 01:18:04 UTC
To play with the new vdpau builds of xbmc make this change in the ebuild

#ESVN_REPO_URI="https://xbmc.svn.sourceforge.net/svnroot/xbmc/branches/linuxport/XBMC"
ESVN_REPO_URI="https://xbmc.svn.sourceforge.net/svnroot/xbmc/branches/xbmc-vdpau/XBMC"

ongoing discussion here - http://xbmc.org/forum/showthread.php?t=45525

testing and working with a nvidia 8500gt, quite nicely too! I renamed it to xbmc-vdpau-9999-r2.ebuild and it needed a "emerge -aC xbmc" first to install with the file collisions.
Comment 162 Peter Fern 2009-02-19 02:16:45 UTC
Created attachment 182495 [details]
live ebuild with additional USE flags

This ebuild:
* Adds 'alsa' and 'pulseaudio' flags to customize audio support.
* Adds 'external-libraries' flag for utilizing external ffmpeg and others.  This functionality is not available at today's date, however it has been merged and reverted a few times, but I left the flag in so it works when the functionality is merged again
* Adds 'pvr' flag for tracking the unified-pvr branch (mutually exclusive with vdpau)
* Adds 'vdpau' flag for tracking the xbmc-vpdau branch (mutually exclusive with pvr)
* Removes spurious ccache flag
* Cleans up some unnecessary eautoreconfs causing compilation errors with recent trunk
Comment 163 Peter Fern 2009-02-19 02:24:33 UTC
Created attachment 182498 [details]
Added warnings for flags tracking alternate branches

Add some warnings for 'pvr' and 'vdpau' flags
Comment 164 Michael Speth 2009-02-19 17:31:28 UTC
I'm getting an error when building with xbmc-9999-r3.ebuild

Here is the error:
---
 Cannot find $EPATCH_SOURCE!  Value for $EPATCH_SOURCE is:
 * 
 *   /usr/local/portage/xbmc-overlay/media-tv/xbmc/files/xbmc-999_readsector.patch
 *   ( xbmc-9999_readsector.patch )
---
Comment 165 Peter Fern 2009-02-19 22:52:08 UTC
Created attachment 182587 [details]
renamed patch

Chuck this in the files dir, sorry
Comment 166 Michael Speth 2009-02-20 15:05:55 UTC
Any ideas for why I'm getting this error?

make: *** [dvdpcodecs] Error 2
 ^[[31;01m*^[[0m
 ^[[31;01m*^[[0m ERROR: media-tv/xbmc-9999-r3 failed.
Comment 167 Pa 2009-02-20 15:41:25 UTC
I get the same error on this new ebuild.

>
(In reply to comment #166)
> Any ideas for why I'm getting this error?
> 
> make: *** [dvdpcodecs] Error 2
>  ^[[31;01m*^[[0m
>  ^[[31;01m*^[[0m ERROR: media-tv/xbmc-9999-r3 failed.
> 

Comment 168 Pa 2009-02-20 19:27:09 UTC
configure: error:

*** [Gentoo] sanity check failed! ***
*** libtool.m4 and ltmain.sh have a version mismatch! ***
*** (libtool.m4 = 1.5.26, ltmain.sh = 1.5.2) ***

Please run:

  libtoolize --copy --force

if appropriate, please contact the maintainer of this
package (or your distribution) for help.

make[2]: *** [config.status] Error 1
make[2]: Leaving directory `/var/tmp/portage/media-tv/xbmc-9999-r3/work/xbmc-9999/xbmc/cores/dvdplayer/Codecs/libmad'
make[1]: *** [libmad] Error 2
Comment 169 Peter Fern 2009-02-20 20:18:19 UTC
Michael, Pa,
Which use flags do you have enabled?

Michael(In reply to comment #167)
> I get the same error on this new ebuild.
> 
> >
> (In reply to comment #166)
> > Any ideas for why I'm getting this error?
> > 
> > make: *** [dvdpcodecs] Error 2
> >  ^[[31;01m*^[[0m
> >  ^[[31;01m*^[[0m ERROR: media-tv/xbmc-9999-r3 failed.
> > 
> 

Comment 170 Michael Speth 2009-02-20 20:42:12 UTC
USE="alsa debug joystick opengl pulseaudio -autostart -external-libraries -gprof -pvr -vdpau" 
Comment 171 Peter Fern 2009-02-20 21:39:00 UTC
Created attachment 182692 [details]
Added some eautoreconfs

Try this, if it doesn't work for you, build with:

MAKEOPTS="-j1" emerge xbmc

And post the full compile error
Comment 172 SpanKY gentoo-dev 2009-02-21 01:42:49 UTC
xbmc-9999.ebuild now in the tree
Comment 173 Brian 2009-02-23 19:59:25 UTC
Getting the following error when building xbmc, not sure if this is the wrong place to post this...

../../lib/libmisc.a(argp-help.o): In function `hol_entry_first_short':
argp-help.c:(.text+0x994): undefined reference to `until_short'
collect2: ld returned 1 exit status
make[5]: *** [rs-euphoria] Error 1
make[5]: Leaving directory `/var/tmp/portage/media-tv/xbmc-9999-r3/work/xbmc-9999/xbmc/screensavers/rsxs-0.9/src/euphoria'
make[4]: *** [all-recursive] Error 1
make[4]: Leaving directory `/var/tmp/portage/media-tv/xbmc-9999-r3/work/xbmc-9999/xbmc/screensavers/rsxs-0.9/src'
make[3]: *** [all-recursive] Error 1
make[3]: Leaving directory `/var/tmp/portage/media-tv/xbmc-9999-r3/work/xbmc-9999/xbmc/screensavers/rsxs-0.9'
make[2]: *** [all] Error 2
make[2]: Leaving directory `/var/tmp/portage/media-tv/xbmc-9999-r3/work/xbmc-9999/xbmc/screensavers/rsxs-0.9'
make[1]: *** [..] Error 2
make[1]: Leaving directory `/var/tmp/portage/media-tv/xbmc-9999-r3/work/xbmc-9999/xbmc/screensavers/rsxs-0.9/xbmc'
make: *** [screensavers] Error 2
 *
 * ERROR: media-tv/xbmc-9999-r3 failed.
 * Call stack:
 *               ebuild.sh, line   49:  Called src_compile
 *             environment, line 3027:  Called die
 * The specific snippet of code:
 *       emake || die "Make failed!";
 *  The die message:
 *   Make failed!

I ran: MAKEOPTS="-j1" emerge -av xbmc
USE="alsa debug opengl -autostart -external-libraries -gprof -joystick -pulseaudio -pvr -vdpau"
Comment 174 SpanKY gentoo-dev 2009-02-23 21:58:06 UTC
this bug is closed.  new issues -> new bugs.

also, the only supported ebuild is the one in the tree.  if you want to use a different one, ask for help on the forums.
Comment 175 Michael Speth 2009-02-24 14:54:47 UTC
Use Flags
media-tv/xbmc-9999-r3  USE="alsa debug joystick opengl pulseaudio -autostart -external-libraries -gprof -pvr -vdpau

ERROR: media-tv/xbmc-9999-r3 failed.
 * Call stack:
 *               ebuild.sh, line   49:  Called src_unpack
 *             environment, line 2995:  Called epatch '/usr/local/portage/xbmc-overlay/media-tv/xbmc/files/xbmc-9999_readsector.patch'
 *             environment, line 1708:  Called die
 * The specific snippet of code:
 *                   die "Failed Patch: ${patchname}!";
 *  The die message:
 *   Failed Patch: xbmc-9999_readsector.patch!
Comment 176 Peter Fern 2009-02-24 23:38:26 UTC
(In reply to comment #175)
> Use Flags
> media-tv/xbmc-9999-r3  USE="alsa debug joystick opengl pulseaudio -autostart
> -external-libraries -gprof -pvr -vdpau
> 
> ERROR: media-tv/xbmc-9999-r3 failed.
>  * Call stack:
>  *               ebuild.sh, line   49:  Called src_unpack
>  *             environment, line 2995:  Called epatch
> '/usr/local/portage/xbmc-overlay/media-tv/xbmc/files/xbmc-9999_readsector.patch'
>  *             environment, line 1708:  Called die
>  * The specific snippet of code:
>  *                   die "Failed Patch: ${patchname}!";
>  *  The die message:
>  *   Failed Patch: xbmc-9999_readsector.patch!
> 

This patch has just been committed to the tree, hence it is no longer required, but as SpanKY says - there is an official ebuild in portage now, so please use that, and post any problems in a new bug, as this one is now closed!