On Dienstag 02 März 2010, Fabrizio Milo aka misto wrote:
> Hi,
> 
> In attachment some patches I made while reading the pycuda code for the
>  git. These are just minor / esthetic fixes.
> 
> Some comments while reading the code:
> 
> in pycuda/driver.py
> What is the motivation to do from _driver import * ?
> 
> I would propose to eliminate the _add_functionality patcher and
> subclass directly the _driver.Device
> something like [snip]
> 
> maybe this can't be done? I am missing something?

Most instances of these classes are created from C++ and hence wouldn't
benefit from addition-by-inheritance. I agree that it's not
super-pretty, but it's necessary.

> I would like also to understand the motivation behind all those white
>  spaces :)

Habit. If you feel you need to 'fix' coding style, there's a matter of
principle here: Either fix the entire code base, or leave it alone. :)

> And why many modules get imported inside the functions instead
>  of
> global scope (for performances maybe ?)

Habit again. Performance-wise, I'm not sure it matters. I like it
because it keeps the global scope free of litter and avoids
long.dotted.names.in.functions.

Regarding your indiviual patches:

1: pure esthetics, meh.
reg super: http://fuhm.net/super-harmful/

2: applied
3: applied, minus style changes
4: you introduced a bug here--do you test your patches?
applied, minus bug

5,6: nope
7: nope, I try to keep the order the same as in the CUDA header, for
easy updating.
8,9: nope

Thanks for your patches.

Andreas

Attachment: signature.asc
Description: This is a digitally signed message part.

_______________________________________________
PyCUDA mailing list
pyc...@host304.hostmonster.com
http://host304.hostmonster.com/mailman/listinfo/pycuda_tiker.net

Reply via email to