[Tutor] Moving my C++ code to Python?

[Sithembewena Lloyd Dube]

Tis a pleasure, Paul. have no experience with Cython vs Boost.Python
benchmarking, but a real world user of the latter had something to say at
http://www.python.org/about/quotes/ (last review).

On Sun, Oct 17, 2010 at 11:29 PM, Paul <paulanon at gmail.com> wrote:

> Thank you all for your kind suggestions.  It seem that cython and
> boost.python both look like a good suggestions; SWIG seems like a good
> tool, but may not meet my requirements.
> Quick google seems to suggest that boost.python is 7 times slower [1],
> however.  Any thoughts about speed performance between the two?
>
> [1]
> http://blog.chrischou.org/2010/02/28/simple-benchmark-between-cython-and-boost-python/
>
> Thank you,
> Paul
>
> And a link I found today.
> [2]
> http://koichitamura.blogspot.com/2008/06/various-ways-to-integrate-python-and-c.html
>
> On Sun, Oct 17, 2010 at 1:56 PM, Sithembewena Lloyd Dube
> <zebra05 at gmail.com> wrote:
> > Also have a look at Boost.Python:
> >
> > http://www.boost.org/doc/libs/1_44_0/libs/python/doc/index.html
> >
> > On Fri, Oct 15, 2010 at 11:29 PM, Paul <paulanon at gmail.com> wrote:
> >>
> >> Hi there,
> >>
> >> I've been using C/C++ for many years (python, just reading about it).
> >>
> >> I have a software written in C/C++ but considering porting most of it
> >> to python, as it seems like it's a better choice for decision making
> >> portion of the code.  I'm also thinking about having a 'matlab' like
> >> interface for reading, processing, and writing.
> >>
> >> In my current C++ code, I would read data into a vector of structs
> >> (which contains other simple vectors, strings, and structs) and it can
> >> be as large as 500MB to 2GB.  The data gets processed (requires random
> >> access).  The result is then written out.
> >>
> >> I would like to make modules for python.  The problem is that the
> >> vector of structs that is very large.  First, is it possible to pass
> >> such structures around to and from python and C/C++?  What would be
> >> the overhead cost of having a large structure that needs to be passed
> >> to and from the C/C++ modules?
> >>
> >> # I imagine I'd use the newly written software this way:
> >> >>> import c_stuff  # my C/C++ module
> >> >>> largestuff = c_stuff.read(file)  # read from disk
> >> >>> c_stuff.process(largestuff, someparams1)  # processing, requires
> >> >>> random access to the vector
> >> >>> c_stuff.process(largestuff, someparams2)
> >> ...
> >> >>> c_stuff.process(largestuff, someparams10000) # the whole thing may
> >> >>> take a few minutes to days
> >> >>>
> >> >>> import python_stuff # some module written in python to process the
> >> >>> data as well
> >> >>>
> >> >>> python_stuff.process(largestuff, otherparams1)  # It's important
> that
> >> >>> this data can be read (and I hope written) by python code
> >> >>> python_stuff.process(largestuff, otherparams2)
> >> >>>
> >> >>> c_stuff.write(largestuff) #write result
> >>
> >> Thank you in advance,
> >> Paul
> >> _______________________________________________
> >> Tutor maillist  -  Tutor at python.org
> >> To unsubscribe or change subscription options:
> >> http://mail.python.org/mailman/listinfo/tutor
> >
> >
> >
> > --
> > Regards,
> > Sithembewena Lloyd Dube
> > http://www.lloyddube.com
> >
>



-- 
Regards,
Sithembewena Lloyd Dube
http://www.lloyddube.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/tutor/attachments/20101018/66d9f301/attachment-0001.html>