Re: [gentoo-user] Re: How serious is revdep-rebuild failure

2005-11-26 Thread Richard Fish
On 11/26/05, Harry Putnam [EMAIL PROTECTED] wrote:
 Harry Putnam [EMAIL PROTECTED] writes:

  Richard Fish [EMAIL PROTECTED] writes:
 
  conftest  echo works
 
  root # ./conftest  echo works
  works
  Seems to have worked as expected.

 Looking at qpkg -v -I|grep gcc
   root # qpkg -v -I|grep gcc
sys-devel/gcc-3.3.5.20050130-r1 *
sys-devel/gcc-3.4.4-r1 *
sys-devel/gcc-config-1.3.12-r4 *

 Is it normal to have 2 versions installed?

Yes.  Gcc is slotted, so it is normal to have more than one version installed.


 Would a more recent version be likely to solve the cross-compiler
 problem?

3.4.4-r1 is the most recent version available for ~x86.

The only thing I can think of that might fix it would be an emerge -e
system.  But could you post the first 200 lines of
/var/tmp/portage/mod_php-4.4.0/work/php-4.4.0/config.log?  Maybe there
is an answer there...

-Richard

-- 
gentoo-user@gentoo.org mailing list



Re: [gentoo-user] Re: How serious is revdep-rebuild failure

2005-11-26 Thread Richard Fish
On 11/26/05, Harry Putnam [EMAIL PROTECTED] wrote:
 Richard Fish [EMAIL PROTECTED] writes:

  /var/tmp/portage/mod_php-4.4.0/work/php-4.4.0/config.log
 Some stuff after 200 lines looks like it might be pertinent so posting
 250 lines.  I hope you see something:

 This file contains any messages produced by compilers while
 running configure, to aid debugging if configure makes a mistake.

 configure:1656: checking host system type
 configure:1756: checking for gcc
 configure:1869: checking whether the C compiler (gcc -O2 -march=pentium4 
 -fomit-frame-pointer  -L/usr/X11R6/lib -ltiff -L/usr/lib) works
 configure:1885: gcc -o conftest -O2 -march=pentium4 -fomit-frame-pointer   
 -L/usr/X11R6/lib -ltiff -L/usr/lib conftest.c  -lxmlparse -lxmltok 15
 /usr/lib/gcc/i686-pc-linux-gnu/3.4.4/../../../../i686-pc-linux-gnu/bin/ld: 
 warning: libmysqlclient.so.12, needed by /usr/X11R6/lib/libxmlparse.so, not 
 found (try using -rpath or -rpath-link)

Ah, here is a problem.  A broken library dependancy.  I don't know
what package libxmlparse is a part of, but it is linked against
libmysqlcliient.so.12, which does not exist now.

Run equery belongs /usr/lib/libxmlparse.so, and rebuild   (with
emerge --oneshot pkg) whatever package that is a part of.

-Richard

-- 
gentoo-user@gentoo.org mailing list



Re: [gentoo-user] Re: How serious is revdep-rebuild failure

2005-11-26 Thread Richard Fish
On 11/26/05, Harry Putnam [EMAIL PROTECTED] wrote:
 Richard Fish [EMAIL PROTECTED] writes:

  Is it normal to have 2 versions installed?
 
  Yes.  Gcc is slotted, so it is normal to have more than one version 
  installed.
 

 Do I need two versions?.

Technically, no.  But this is where I get a bit paranoid, because if
you break the libstdc++ dynamic linking, you cripple the system
(including portage!).

So, I would only prune the old versions of gcc after doing a emerge
-e world to rebuild *everything* with the new gcc.

Oh, and a fix_libtool_files.sh also...

And making a good backup...

-Richard

-- 
gentoo-user@gentoo.org mailing list



Re: [gentoo-user] Re: How serious is revdep-rebuild failure

2005-11-25 Thread Holly Bostick
Harry Putnam schreef:

 First let me add that (mjpegtools-1.6.2-r3) isn't even installed: 
 root # qpkg -v -I |grep mjpegtools media-video/mjpegtools-1.8.0-r1 *
 
 
 equery depends mjpegtools [ Searching for packages depending on 
 mjpegtools... ] media-video/transcode-0.6.14-r2 
 media-video/cinelerra-cvs-20050801
 
 both are installed here: # qpkg -v -I |grep 'transcode\|cinelerra' 
 media-video/transcode-0.6.14-r2 * media-video/cinelerra-cvs-20050801
  *
 

This seems odd:

Runtime Dependencies
transcode-0.6.14-r2

media-libs/libexif
media-libs/netpbm
|   = media-video/ffmpeg - 0.4.9_pre1
a52 = media-libs/a52dec - 0.7.4
avi = media-video/avifile - 0.7.41.20041001
dv = media-libs/libdv - 0.99
dvdread = media-libs/libdvdread - 0.9.0
encode = media-sound/lame - 3.93
fame = media-libs/libfame - 0.9.1
gtk = x11-libs/gtk+ - 1.2*
imagemagick = media-gfx/imagemagick - 5.5.6.0
jpeg media-libs/jpeg
lzo = dev-libs/lzo - 1.08
mjpeg = media-video/mjpegtools - 1.6.2-r3
mpeg media-libs/libmpeg3
ogg media-libs/libogg
pvm = sys-cluster/pvm - 3.4
quicktime = media-libs/libquicktime - 0.9.3
sdl media-libs/libsdl
theora media-libs/libtheora
truetype = media-libs/freetype - 2
vorbis media-libs/libvorbis
xvid = media-libs/xvid - 1.0.2
X virtual/x11

Runtime Dependencies
cinelerra-cvs-20050801

media-libs/faad2
media-libs/libdv
|   = media-libs/libogg - 1.0
media-libs/libpng
|   = media-libs/libtheora - 1.0_alpha4-r1
|   = media-libs/libvorbis - 1.0.1-r2
|   = media-libs/openexr - 1.2.1
|   = media-sound/esound - 0.2.34
! media-video/cinelerra -
! media-video/cinelerra -
media-video/mjpegtools
|   = sys-libs/libavc1394 - 0.4.1
|= sys-libs/libraw1394 - 0.9.0
virtual/x11

What seems odd to me is that neither of these two programs depends on
that specific version of mpegtools. Transcode will take that version or
above, cinelerra doesn't care.

And, since you have a greater version installed, there seems to be no
reason that revdep-rebuild should be trying to install an older version
in this way.

The highest likelihood is that-- as previously suggested-- you're
running an old output from revdep-rebuild -p (when this version was the
installed version of mjpegtools), and did not delete the old output
files and re-run revdep-rebuild -p to get a new prospective rebuild
list. You might want to check your /root/ folder and see what the dates
on those .revdep-rebuild files is. If they aren't from a recent time,
remove them, and run revdep-rebuild (-p) again.

Alternatively (hacky solution), try re-emerging cinelerra and transcode
again (to retrain them in where their dependent files are and what
version they are).

Even more hackily, remove mjpegtools totaly (unmerge), then do an emerge
-uaDtv world, and let it get pulled back in as a dependency of the two
packages that need it. That ought to straighten everything out.

But probably you're just using old revdep-rebuild output, and the
easiest solution would be to delete those files so that revdep-rebuild
can update itself with the current state of your system.

HTH,
Holly

-- 
gentoo-user@gentoo.org mailing list



Re: [gentoo-user] Re: How serious is revdep-rebuild failure

2005-11-25 Thread Holly Bostick
Harry Putnam schreef:
 Richard Fish [EMAIL PROTECTED] writes:
 
 (Including Richard in reply as well) Nagatoro [EMAIL PROTECTED] 
 writes:
 [...]
 
 Assigning files to ebuilds... using existing 
 /root/.revdep-rebuild.4_ebuilds. Evaluating package order... 
 using existing
 
 Nagatoro replied:
 
 ^^^using existing^^^ means that you are using the results from an 
 older run of revdep-rebuild. First remove all .revdep* files and 
 the run it again and see if you find the same errors.
 
 Harry responds:
 
 Ack, yes of course and it even warns you about that
 
 However having removed them I still get a huge list of stuff listed 
 as BROKEN

Yes, well, that's what revdep-rebuild does-- finds broken stuff. It's
doing its job-- what's the problem with that?
 
 One of the first involves the same mjpeg package that isn't even 
 installed:
 
 broken /usr/bin/cinelerra (requires  libmjpegutils-1.6.so.0)


U--- why do you feel that this is the same package that isn't even
installed? You said that you have libmjpegutils installed, just not the
same version that was attempting to be rebuilt before 1.8.0 installed,
rather than the 1.6.2-r3 that was attempting to be rebuilt).

 eix mjpegtools
* media-video/mjpegtools
 Available versions:  1.6.2-r4 ~1.8.0 ~1.8.0-r1
 Installed:   1.6.2-r4
 Homepage:http://mjpeg.sourceforge.net/
 Description: Tools for MJPEG video

equery files media-video/mjpegtools
[ Searching for packages matching media-video/mjpegtools... ]
* Contents of media-video/mjpegtools-1.6.2-r4:

/usr/lib/libmjpegutils-1.6.so.0 - libmjpegutils-1.6.so.0.2.2

Now, obviously this is not the same version of mjpegtools that you have,
but what it indicates is that the file libmjpegutils-1.6.so.0 is a
symlink to whatever version of the actual library is installed by the
package.

I rather expect that what would happen if I were to upgrade this package
is that the symlink itself would remain, but the target of the symlink
would change.

If this is in fact the case, two points:

1: your symlink seems to be broken;
2: the error you have listed does not say anything about what version of
mjpegtools is installed or broken, so revdep-rebuild is not necessarily
talking about the same version as previously,


But better to go to the source:
 
 == Full output of revdep-rebuild 
 (minus all config make stuff [sorry about control chars I forgot to 
 use -nc but have removed some]):
 
 Note it doesn't appear to say what pkg actually failed:
 
 
 Evaluating package order... done. (/root/.revdep-rebuild.5_order)
 
 All prepared. Starting rebuild.. emerge --oneshot 
 =dev-php/mod_php-4.4.0 =dev-php/php-4.4.0 =media-libs/imlib-1.9.14-r3
  =kde-base/kdegraphics-3.4.1-r1 =media-gfx/imagemagick-6.2.5.4 
 =media-libs/libdv-0.102 =media-video/avifile-0.7.41.20041001-r1 
 =media-video/cinelerra-cvs-20050801 =media-video/transcode-0.6.14-r2
  =net-libs/libwww-5.4.0-r3 ^G.^G.^G.^G.^G.^G.^G.^G.^G.^G.
 
 --8 [big snip] 
 
 you have the following choices:
 
 - if emerge failed during the build, fix the problems and re-run 
 revdep-rebuild

So apparently the rebuild failed.

But first of all, I don't see mjpegtools being rebuilt in this list, so
that is not the problem apparently (the problem is not that mjpegtools
is not installed, but that the programs that depend on it are not linked
against it, which is what revdep-rebuild is trying to fix by re-emerging
them);

... and second of all, which package failed to emerge and why?

Meaning, what was the error in whichever package failed to emerge?

Holly
-- 
gentoo-user@gentoo.org mailing list



Re: [gentoo-user] Re: How serious is revdep-rebuild failure

2005-11-25 Thread Richard Fish
On 11/25/05, Harry Putnam [EMAIL PROTECTED] wrote:
 It turn out that using old revdep output was not the problem.  See
 just posted output in response to Nagatoro.

Actually it was.  Notice that revdep-rebuild is no longer trying to
rebuild mjpegtools, but those things that depend upon mjpegtools
(along with other broken packages).

So, what package is actually failing to merge, and what is the error
message that is being given?

-Richard

-- 
gentoo-user@gentoo.org mailing list



Re: [gentoo-user] Re: How serious is revdep-rebuild failure

2005-11-25 Thread Richard Fish
On 11/25/05, Harry Putnam [EMAIL PROTECTED] wrote:
 Harry Putnam [EMAIL PROTECTED] writes:

  Oh crap.. overzealous snippage caused me to leave out the main stuff:

 I seem to have taken a moron pill this morning please see full output
 of revdep-rebuild in a few minutes at:
 http://www.jtan.com/~reader/vu_txt/display.shtml (in 5 min or so)

Ok, now we are going to need to see the output of emerge --info,
because for some reason your toolchain thinks it is cross-compiling:

 checking whether the C compiler (gcc -O2 -march=pentium4 -fomit-frame-pointer 
  -L/usr/X11R6/lib -ltiff -L/usr/lib) works... yes
 checking whether the C compiler (gcc -O2 -march=pentium4 -fomit-frame-pointer 
  -L/usr/X11R6/lib -ltiff -L/usr/lib) is a cross-compiler... yes

-Richard

-- 
gentoo-user@gentoo.org mailing list



Re: [gentoo-user] Re: How serious is revdep-rebuild failure

2005-11-25 Thread Holly Bostick
Harry Putnam schreef:
 
 I still don't see the actual error there and it was the output of:
 
 revdep-rebuild -nc  21|tee revdep.log
 
 revdep.log is what I posted online.
 

Richard Fish replied with the specific issue about half an hour ago:

 Richard Fish schreef:
 
 Ok, now we are going to need to see the output of emerge --info, 
 because for some reason your toolchain thinks it is cross-compiling:
 
 
 checking whether the C compiler (gcc -O2 -march=pentium4 
 -fomit-frame-pointer  -L/usr/X11R6/lib -ltiff -L/usr/lib) works...
  yes checking whether the C compiler (gcc -O2 -march=pentium4 
 -fomit-frame-pointer  -L/usr/X11R6/lib -ltiff -L/usr/lib) is a 
 cross-compiler... yes
 
 

The problem occurs with the very first compile

configure: error: can not run test program while cross compiling
  ...done!
| emerge (1 of 10) dev-php/mod_php-4.4.0 to /

... so the problem is definitely somewhere in your toolchain, rather
than anything to do with either revdep-rebuild or the specific programs
attempting to be rebuilt.

Posting emerge info is a good starting point to troubleshoot this
(unless you already happen to know why this is occurring, that's also
possible).

Holly
-- 
gentoo-user@gentoo.org mailing list



Re: [gentoo-user] Re: How serious is revdep-rebuild failure

2005-11-25 Thread Richard Fish
On 11/25/05, Harry Putnam [EMAIL PROTECTED] wrote:
 Richard Fish [EMAIL PROTECTED] writes:

  Ok, now we are going to need to see the output of emerge --info,
  because for some reason your toolchain thinks it is cross-compiling:

 There appears to be some confusion in that output as to what USE flags
 are in force.
ACCEPT_KEYWORDS=x86 ~x86
 I guess the last is what is used?

 I have in /etc/make.conf

 ACCEPT_KEYWORDS=~x86
 This is because a few -u worls back (2 I think) I foolishly ran
 ACCEPT_KEYWORDS='~x86' emerges  -v -u -D world

Well, the only way ~x86 could have been added to make.conf was if it
was edited directly.  Running with ACCEPT_KEYWORDS in the environment
would not have changed it.

Having --info report both x86 and ~x86 is normal...it means both
stable and testing packages are allowed.


 And it seems more trouble to back out of that now than to just try to
 see if I (with help) can stay on top of it.

Well, it isn't terribly difficult to switch back to stable.  Just
remove the ~x86 keyword from make.conf and emerge -DNuv world.  The
worst case is when you have some testing package merged for which
there is no stable version, in which case you either have to unmerge
the package or add the appropriate entry to
/etc/portage/package.keywords.

And of course, don't forget etc-update afterwards.  Anyway, on the to
the problem at hand:

 emerge -n -v --info 21|tee emergeInfo.log

 Gentoo Base System version 1.6.13

Hmm, I would have expected something more recent for ~x86.  What does
emerge -vp sys-apps/baselayout report?

 Portage 2.0.53_rc7 (default-linux/x86/2005.0, gcc-3.3.5-20050130, 
 glibc-2.3.5-r3, 2.6.14-gentoo-r2 i686)

Ok, run gcc-config -l.  You should see an entry for
i686-pc-linux-gnu-3.4.4.  If so, run gcc-config
i686-pc-linux-gnu-3.4.4  Then try the revdep-rebuld again.

Everything else looks sane.

-Richard

-- 
gentoo-user@gentoo.org mailing list



Re: [gentoo-user] Re: How serious is revdep-rebuild failure

2005-11-25 Thread Richard Fish
On 11/25/05, Harry Putnam [EMAIL PROTECTED] wrote:
 Harry Putnam [EMAIL PROTECTED] writes:

 [...]

  checking whether the C compiler (gcc -O2 -march=pentium4 
  -fomit-frame-pointer  -L/usr/X11R6/lib -ltiff -L/usr/lib) is a 
  cross-compiler... yes

 Still thinks its a cross-compiler... what does that mean anyway?

Well, generally, it means that you are compiling for a different
architecture than what you are running.  For example, it is possible
to compile for AMD64 CPUs from a P4 host, or vice-versa.  However,
since you are not doing that, it means your toolchain is broken in
some way.

The way autoconf (the ./configure script) checks for this is that it
compiles a very simple program.  This program is:


#line 1880 configure
#include confdefs.h

main(){return(0);}

It then compiles this program.  If the program compiles, configure
decides that gcc works.  If the program doesn't run, it decides that
you are cross compiling.  So, let's try this manually.

Save the above lines to a file, call it conftest.c.  The build it with

gcc -O2 -march=pentium4 -fomit-frame-pointer  -L/usr/X11R6/lib -ltiff
-L/usr/lib -o conftest conftest.c

If the compile complets without error, try running the program with:

conftest  echo works

-Richard

-- 
gentoo-user@gentoo.org mailing list



Re: [gentoo-user] Re: How serious is revdep-rebuild failure

2005-11-25 Thread Richard Fish
On 11/25/05, Harry Putnam [EMAIL PROTECTED] wrote:
 Richard Fish [EMAIL PROTECTED] writes:

  #line 1880 configure
  #include confdefs.h
 
  main(){return(0);}
 
  It then compiles this program.  If the program compiles, configure
  decides that gcc works.  If the program doesn't run, it decides that
  you are cross compiling.  So, let's try this manually.
 
  Save the above lines to a file, call it conftest.c.  The build it with
 
  gcc -O2 -march=pentium4 -fomit-frame-pointer  -L/usr/X11R6/lib -ltiff
  -L/usr/lib -o conftest conftest.c

  gcc -O2 -march=pentium4 -fomit-frame-pointer  -L/usr/X11R6/lib \
   -ltiff -L/usr/lib -o conftest conftest.c
   configure:1880:22: confdefs.h: No such file or directory

 Just complains about not finding confdefs.h.

Sorry, my fault.  Confdefs.h is generated by the configure script, and
doesn't matter for this case.  Just remove the include from the file.

-Richard

-- 
gentoo-user@gentoo.org mailing list



Re: [gentoo-user] Re: How serious is revdep-rebuild failure

2005-11-24 Thread Richard Fish
On 11/24/05, Harry Putnam [EMAIL PROTECTED] wrote:
 Richard Fish [EMAIL PROTECTED] writes:

  My guess is 'emerge -u --oneshot mjpegtools' will fix the problem.

 No, it didn't change a thing.  But there was some output at the end
 that might mean something:

 [...]
  * Please upgrade your package (mjpegtools-1.6.2-r3) to use 
 toolchain-funcs.eclass

Interesting.  All 3 builds currently in portage (1.6.2-r4, 1.8.0, and
1.8.0-r1) use toolchain-funcs already.

What is the result of

ls /usr/portage/media-video/jpegtools/*.ebuild

and

equery depends mjpegtools

-Richard

-- 
gentoo-user@gentoo.org mailing list



Re: [gentoo-user] Re: How serious is revdep-rebuild failure

2005-11-24 Thread Richard Fish
On 11/25/05, Richard Fish [EMAIL PROTECTED] wrote:
 What is the result of

 ls /usr/portage/media-video/jpegtools/*.ebuild

 and

 equery depends mjpegtools

Also, do you have anything for mjpegtools in /etc/portage/package.mask?

-Richard

-- 
gentoo-user@gentoo.org mailing list