[AstroPy] Astropy 0.2.5 compilation error using Intel compilers under Mac OS Mavericks and python under Homebrew

[Michael Droettboom]

On 11/18/2013 09:46 AM, Jean-Baptiste Marquette wrote:
>
> Hi again,
>
>> If the latter, then we should open a general issue to support the
>> intel C compiler (I think no one had requested it so far, hence why it
>> went unnoticed). Try setting your compiler to clang instead to see if
>> it helps.
>
> clang successful, thus we are facing an icc compiler issue. Supporting 
> it would be great. Emmanuel Bertin told me it is much more efficient 
> in some situations, for example with sine or cosine functions.

We'll need some way to provide the financial support for this, of 
course.  ICC is far from free.  It doesn't appear on paper that STScI 
would be eligible for academic pricing or non-commercial pricing by 
Intel's definition [1] [2], so it's $699/box or $2,499 for 2-concurrent 
floating site license.  It might be worth experimenting with the 30-day 
evaluation license to see what the performance benefits are before 
committing to long-term support of that compiler.  That kind of money 
actually goes a real long way toward hosted testing solutions etc. which 
would benefit a lot more users than supporting a commercial compiler.  
All that said, we can always accept small patches to support ICC even if 
none of the core developers have a copy.

[1] " For faculty or staff at degree-granting institutions of higher 
learning"
[2] " Non-commercial software development means you are not paid and/or 
compensated in any form, by anyone, for software development using the 
Intel® Software Development Products under the terms of the 
non-commercial license."

Mike

-- 
                    _
|\/|o _|_  _. _ | | \.__  __|__|_|_  _  _ ._ _
|  ||(_| |(_|(/_| |_/|(_)(/_|_ |_|_)(_)(_)| | |

http://www.droettboom.com

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/astropy/attachments/20131119/319baa35/attachment.html>