libvpx.so.0 fails to load on low-memory machines

[Eric Fernandez]

Hi,

I have the same error than previously reported on RedHat and Debian
mailing lists:
https://bugzilla.redhat.com/show_bug.cgi?id=600663
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=583765

This issue was supposed to be corrected with this patch on the 11-06-2010:
https://review.webmproject.org/gitweb?p=libvpx.git;a=commitdiff;h=9099fc0d69a69525656b4bfeeb1e7aabec04897b
However I can reproduce the issue on Arch Linux with the git version
dated 15-06-2010.

Disabling PSNR during build, however, corrects the issue.

Regards,
Eric

[John Koleszar]

--enable-psnr should never be set by distributions. its purpose is to
enable creation of a file 'opsnr.stt' which has some psnr and ssim
numbers in it. This file will be dropped in the current directory
unconditionally, which is not the behavior you want from a lib
installed system-wide. Users interested in getting PSNR information
from the encoder should use the PNSR packet interface (enabled with
the VPX_CODEC_USE_PSNR flag) which does not depend on --enable-psnr
being set.

I looked at the arch linux build spec[1], and other than the log entry
indicating removing this option was temporary, it looks ok to me.

http://repos.archlinux.org/wsvn/packages/libvpx/trunk/PKGBUILD?op=log&

[Eric Fernandez]

2010/6/16 John Koleszar <jkol...@google.com>:

>
> --enable-psnr should never be set by distributions. its purpose is to
> enable creation of a file 'opsnr.stt' which has some psnr and ssim
> numbers in it. This file will be dropped in the current directory
> unconditionally, which is not the behavior you want from a lib
> installed system-wide. Users interested in getting PSNR information
> from the encoder should use the PNSR packet interface (enabled with
> the VPX_CODEC_USE_PSNR flag) which does not depend on --enable-psnr
> being set.
>
> I looked at the arch linux build spec[1], and other than the log entry
> indicating removing this option was temporary, it looks ok to me.
>
> http://repos.archlinux.org/wsvn/packages/libvpx/trunk/PKGBUILD?op=log&
>

Hi John,

Thanks a lot for your answer. I am in contact with the PKGBUILD author
and he decided to remove the PSNR option after my bug report. Thanks a
lot for your info about PSNR information, it was not clear if
--enable-psnr was required to get the PSNR information. Maybe this
option could be renamed ?

[Luca Barbato]

On 06/16/2010 02:47 PM, Eric Fernandez wrote:
> Thanks a lot for your answer. I am in contact with the PKGBUILD author
> and he decided to remove the PSNR option after my bug report. Thanks a
> lot for your info about PSNR information, it was not clear if
> --enable-psnr was required to get the PSNR information. Maybe this
> option could be renamed ?

It is being evaluated here:

http://code.google.com/p/webm/issues/detail?id=35

you might want to track it as well =)

lu

--

Luca Barbato
Gentoo/linux
http://dev.gentoo.org/~lu_zero