Hi Ido and Carlos,

can you check if that tarball is working?

http://downloads.sourceforge.net/project/math-atlas/Developer%20%28unstable%29/3.11.11/atlas3.11.11.tar.bz2

The chance is low, but if its working for you I will consider to create a new version for it.
Thanks,
Bjoern

Yes this tar is broken at least on OSX. 
Other people have the same issue:
http://code.google.com/p/libarchive/issues/detail?id=299



On Sep 20, 2013, at 10:41 AM, Bjoern Gruening <bjoern.gruening@gmail.com> wrote:

> Hi Carlos,
> 
>>> Hi Peter and Carlos,
>>> 
>>>> On Mon, Sep 16, 2013 at 8:57 PM, Carlos Borroto
>>>> <carlos.borroto@gmail.com> wrote:
>>>>> I did an extra test. Started with a clean 'galaxy-dist'. This time
>>>>> both repositories fail with the same error. I guess before something
>>>>> was cached for the repository with version 0.1.4.
>>>>> 
>>>>> I used biopython repository as a guide to write my tool dependency definition:
>>>>> http://testtoolshed.g2.bx.psu.edu/view/biopython/package_biopython_1_61
>>>>> 
>>>>> I can confirm biopython repository is failing to install for me with
>>>>> exactly the same error. I wonder if a recently addition in the test
>>>>> toolshed broke the treatment of "prior_installation_required".
>>>>> 
>>>>> Thanks,
>>>>> Carlos
>>>> 
>>>> Could be - note that currently Biopython isn't currently
>>>> installing properly on the Test Tool Shed due to ALTAS
>>>> failing (a requirement of NumPy which is a requirement
>>>> of Biopython). Dave and Bjoern are I think looking at this
>>>> already...
>>> 
>>> I can't do much I tested it again and for me its working fine on my
>>> computers I have at hand ... sorry.
>>> 
>> 
>> In case it helps, this is how the INSTALLATION.log file ends on OS X 10.8.4:
>> $ tail -n 3 ~/src/shed_tools_dependencies.central/atlas/3.10.1/iuc/package_atlas_3_10/3508de0ebae1/INSTALLATION.log
>> x ATLAS/tune/lapack/lanbsrch.c
>> tar: Error exit delayed from previous errors.
>> #############################################
>> 
>> This is the relevant part I can find in Galaxy's log:
>> [localhost] local: tar xfvj atlas3.10.1.tar.bz2
>> 
>> Warning: local() encountered an error (return code 1) while executing
>> 'tar xfvj atlas3.10.1.tar.bz2'
>> 
>> After noticing this I got what I'm guessing is the original file from
>> sourceforge:
>> http://downloads.sourceforge.net/project/math-atlas/Stable/3.10.1/atlas3.10.1.tar.bz2
>> 
>> I can confirm that trying to untar this file fails with the exact same
>> error. However, on Ubuntu 13.04 untaring this file works just fine.
> 
> That is new to me. How can that happen? Anyone with an OS-X can confirm
> that?
> 
>> On Ubuntu 13.04 the error I see is:
>> STDERR
>> It appears you have cpu throttling enabled, which makes timings
>> unreliable and an ATLAS install nonsensical.  Aborting.
>> See ATLAS/INSTALL.txt for further information
>> #############################################
>> 
>> Björn, you said cpu throttling can be easily disable on Ubuntu. Can
>> you comment how? Do I need to disable it completely, or should the
>> install script do it when installing?
> 
> ATLAS (once untared ;-)) needs cpu-throttling to be disabled to
> optimizes its library. If it is not disabled ATLAS compilation will
> fail. For OS-X I found that:
> 
> http://apple.stackexchange.com/questions/41045/how-can-i-disable-cpu-throttling-and-cpu-disabling
> 
> Sorry, I never touched a OS-X. Nevertheless, If its not disabled, it is
> supposed to fail silently and downstream packages will not be affected.
> But if its crashing during untarring, I can't do much. What I can do is
> to repack the tarball and host it elsewhere. What brings me to:
> 
> http://gmod.827538.n3.nabble.com/RFC-Storing-of-tarballs-and-patches-for-tool-dependencies-to-enable-reproducibility-td4036591.html
> 
> Bad news for a Friday morning, sorry :(
> Bjoern
> 
> 
>> Thanks,
>> Carlos
>> 
>> ___________________________________________________________
>> Please keep all replies on the list by using "reply all"
>> in your mail client.  To manage your subscriptions to this
>> and other Galaxy lists, please use the interface at:
>>  http://lists.bx.psu.edu/
>> 
>> To search Galaxy mailing lists use the unified search at:
>>  http://galaxyproject.org/search/mailinglists/
> 
> 
> 
> ___________________________________________________________
> Please keep all replies on the list by using "reply all"
> in your mail client.  To manage your subscriptions to this
> and other Galaxy lists, please use the interface at:
>  http://lists.bx.psu.edu/
> 
> To search Galaxy mailing lists use the unified search at:
>  http://galaxyproject.org/search/mailinglists/