[lsst-dm-stack-users] Trying out the Anaconda python with a RHEL6 installation

Heather Kelly heather at slac.stanford.edu
Fri Feb 7 09:23:37 PST 2014



On 2/7/2014 12:14 PM, Chris Walter wrote:
> Hi Heather,
>
>> If I do:
>> bash-4.1$ conda list numpy
>> # packages in environment at
>> /astro/astronfs01/workarea/lsst/software/anaconda/1.8.0:
>> #
>> numpy                     1.7.1                    py27_0
>>
>>
>> I'm trying Winter 2013 (as that has been the version we've stuck with
>> for awhile)..and it does seem to be set up to use numpy 1.6.  I'm
>> thinking I need to move to Summer2013...   I'll try that
>> now and cross my fingers.
> Hmm... OK, well it sounds like you accidentally still built numpy when you built Winter2013.  Check that you did the manifest remap steps correctly.  If you do it right you won't have any numpy1.6 for eups to load.
>
> Good luck,
>
> -Chris
>
>
>

Actually skipping the manifest remap step was by design :)  At least
that's how I read Mario's last message from 2/2/2014:


> Heather, try the following: start from scratch. Make sure you have
> anaconda in path. Follow the instructions, but after you download
> newinstall.sh, edit it and change the $(uname) == Darwin test in:
>
> ========
> ...
> # Mac-specific setup: mock python and gcc packages to satisfy dependencies;
> # system python will be used. Note: gcc is not supported on OSX; use clang.
> #
> [ "$(uname)" == "Darwin" ] && {
>     echo "Detected Mac OS X; will use system-provided python."
> ...
> ========
>
> to $(uname) == Linux (so it will use system-provided python on Linux as
> well). Then you don't need to manually do the manifest manipulation I
> suggested in my original e-mail.
>
> I think that may fix it.

The manifest.remap bit probably will fix it..but regardless..  I think
it's a waste of DM's (and my) time to play with the older release.  I'll
move on to Summer2013.

Take care,
Heather



More information about the dm-users mailing list