yade-users team mailing list archive

[Václav Šmilauer <eudoxos@xxxxxxxx>]

>    From my experience : I feeled that the time I spent to learn a bit 
> C++ was not lost : it helped me to find some known structures and 
> characters in the ocean of code I had in front of me at the 
> beginnings. Quite good when you have nothing to hang at...
>
>    For the other things, I'm very not up to date (and the activity of 
> these last days let me think that  this will be incredibly worser and 
> worser) with the latest Yade facilities so I'm not sure my opinion is 
> to be considered. But it seems to me that Python is not too much 
> complicated, taking moreover into account the fact that some examples 
> exist (and that Vaclav always answer rather quickly to such questions 
> on the list, thanks to him !!)
>    And I do not even know what is this MatPlotlib ?? I guess it is for 
> plotting results. In my case, I use always Recorder-issued files and 
> Matlab. I find it satisfying.
Never heard of Matlab, where can I download it? ;-)

FYI matplotlib is python plotting module with syntax very similar to 
matlab's: http://matplotlib.sf.net. It is used by the yade.plot module 
and I use it quite a lot for post-processing as well. The beauty of it 
is that you avoid dealing with recorders&files all the time, as you have 
access to everything you need from within python -- all bodies' 
variables, engines, interactions etc.

Concerning c++: it is certainly a big advantage to know the c++ 
internals of yade, but I believe we got to the point that one doesn't 
need to touch it most the time. I do most my work in python these days 
and I get to c++ only to fix bugs or add substantial features.

I would even suggest to rewrite all FileGenerators in python (nice task 
for a beginner as well), to get rid of the idea that you need c++ for 
anything but the weight-lifting during computation. Compare ClumpTestGen 
and scripts/test/clump.py (which are functionally identical) to get the 
difference. I can imagine writing beginner's documentation about 
clump.py, but not about ClumpTestGen. Why should material researchers 
care about Omega or shared_ptr or boost?

V.