[pythran] Re: Preparing PyDATA

  • From: Pierrick Brunet <pierrick.brunet@xxxxxxxx>
  • To: pythran@xxxxxxxxxxxxx
  • Date: Sat, 12 Apr 2014 00:25:43 +0200


I admit that I don't understand. #251 doesn't exists :-(

About import, I think it will not be fully ready in time (Medhi?) so ok for next release. But about globals, I fear it will never be in the master. It becomes more and more obsolete and coyotte doesn't look to have a lot of time for it. The more we wait, the harder it will be to integrate. I think it should really be in this version.

About test cases, are you thinking about perf? Or supported code? from stack-overflow? Or rosetta? Or both?

On 11/04/2014 23:11, serge Guelton wrote:
Hi Pythraners,

Mehdi is going to present Pythran at PyDATA the 2-4 may 2014. We have to
support him with all our strength, which basically mean releasing a
stable version of Pythran soon.

Concerning on-going PR, it means #241 could make its way in, but maybe
none of the others. It also mean I have to implement OpenMP
parallelisation of numpy operations before reviewing global variables. I
quite don't like the way #251 is implemented, but I miss the time for a
deep inspection, so I don't know about this one. #242 introduces a major
new feature, so I'd rather wait for this one to be fully tested on a lot
of test cases too.

So basically, my *proposal* is:

- before the release: #241 & # 251 & #OpenMP + more test cases on the
   numpy part (I strongly encourage pbrunet to extract test cases from
   stack overflow, and I'll do it too!)

- after the release: keep on having fun with Pythran!

any thought?

Other related posts: