PythonOgre Development

44 views
Skip to first unread message

Andy Miller

unread,
Aug 18, 2010, 12:34:16 AM8/18/10
to python-ogre-developers
Team

As you've no doubt noticed PythonOgre development has stabilised (or slowed to a crawl based upon it being fairly complete :) )

I've been looking at future projects and wanted a little feedback/debate on a couple to see if there is any interest (any thoughts would be welcome) -- and of course please add anything else you see fit to this thread....

'Faster Python'
1. ShedSkin:  A nice python to C++ 'compiler' - I helped out with the MSVC support and have been thinking about how best to link PythonOgre modules with it -- challenge is that it's unlikely to support a "complete" python app due to a number of limitations -- however possibly ideal to create 'speed-up' modules.
2. PYPY: More of a JIT compiler which recently added a 'ctypes' style interface -- I've done initial work to show that Boost can be compiled against it and now looking at testing a module or two (OIS for example)

'Something Different with GO'
The GO language from google looks rather interesting, admittedly it's intended for systems programing (which I like) so more complex than Python (perhaps) -- for me it fills a middle ground between 'C' (which at times is just too much work), Java (which I've never liked due to performance, portability and the size of the library documentation -- much of which is getting better), C++ (which I like 60% of but complex templates, STL, and debugging multiple inheritance drive me nuts) and Python (which I'd like to see as a compiled language, but appreciate the fantastic out of the box usability)

Of course GO is lacking as there is currently no C++ interface or ability to call existing C++ libraries, and the discussion I've seen is that SWIG will be the wrapper of choice which is concerning from an Ogre perspective -- yes there is a 'C' interface wrapper and from what I see this is currently manual and not sure about the performance... 

Would appreciate thoughts on Shedskin/Pypy/Go/etc, other random thoughts on PythonOgre development in general, etc

Thanks
Andy

Jeremy Sandell

unread,
Aug 27, 2010, 7:40:25 PM8/27/10
to Python Ogre Developers


On Aug 18, 12:34 am, Andy Miller <nzmill...@gmail.com> wrote:>
> *'Faster Python'*
> 1. *ShedSkin*:  A nice python to C++ 'compiler' - I helped out with the MSVC
> support and have been thinking about how best to link PythonOgre modules
> with it -- challenge is that it's unlikely to support a "complete" python
> app due to a number of limitations -- however possibly ideal to create
> 'speed-up' modules.

I was pretty impressed with how much they already support, though I
did notice it doesn't handle Python 3 / 2.6 string formating, as of
yet.
Along the same lines, I'd recommend looking at Cython, as well. They
released a version this month that improved the C++ support.

> 2. *PYPY*: More of a JIT compiler which recently added a 'ctypes' style
> interface -- I've done initial work to show that Boost can be compiled
> against it and now looking at testing a module or two (OIS for example)

I would definitely be interested in hearing the results. While not on
the same level as, say, LuaJIT, PyPy's speed improvements are quite
impressive.

> and Python (which I'd like to see as a compiled language, but appreciate the fantastic out of the box usability)

Definitely agree about wanting to see a compiled version. Maybe one
day.

Best regards,
Jeremy Sandell

Joe Leaver

unread,
Aug 28, 2010, 1:32:58 PM8/28/10
to Python Ogre Developers
PYPY or Cpython both sounds like a good direction to me. Faster python
is better python.

For something completely different, though I don't think it's quite
ready yet -- why not javascript? Like google V8: http://code.google.com/p/v8/

I don't think C++ support is there yet, but I know it's on the agenda,
and there's been a lot of commercial support for making javascript run
faaast.

Of course, I'm not a huge fan of javascript. I'm happy sticking to
python, for now. I just hope the various projects for python JIT and
optimizations stay alive and keep going.
Reply all
Reply to author
Forward
0 new messages