Thanks for your kind feedback.
On Wed, Aug 26, 2009 at 10:36 AM, Jimmy<jimmy....@gmail.com> wrote:
>
> First and mostly I want to congratulate the developer of this
> impressive work.
> Visualizations from the website are very nice!
>
> Was Summon implied in this?
> http://cs.jhu.edu/~razvanm/fs-expedition/
> If yes, that would be a nice display of what can be achieved.
That looks like an independent project. I don't think it uses SUMMON.
> On to questions, now...
> While on linux, I add no problems installing summon (using
> easy_install), it's much tougher on Windows...where even the binary is
> failing...dont know if you want some feedback on that....
Do you use python2.5? That is the only install that I have made.
> The website states that summon is faster than regular python openGL
> bindings. I assume that you are talking about PyOpenGL and pyglet...
> I've been using pyglet up to now to do some visualisation on data
> matrices and plotting work (to be included in a stand alone app)...do
> you have some rough estimates of the gain that summon is bringing on
> the table compared to these libs?
I do not made a time comparison with those libraries. But python code
in general runs 50 times slower than C code. The biggest slow down
comes in tight loops. The draw function that creates each frame often
needs to loop through all objects to layout/draw. This can be a very
slow loop in python, however, in SUMMON this is done is C++. If you
rely on features such as drawlists and other ways of pushing the loop
into the GPU/native lib you can also get comparable speed ups
(although with additional restrictions on what you can draw) while
coding in python.
> Lastly, I cant make the example: Animation run...
> the timer function is not recognized...
> it says "module object (summon) has no attribute add_update_func...
Hmm. Let me look into that. My current development version does not
have this bug, but perhaps the version on the site does.
Thanks for your feedback.
Matt