[Baypiggies] FW: Speaker for Feb/Mar/Apr [Pyrex/ctypes/SWIG for wrapping image analysis ]

[Stephen McInerney]

Please all change the title to just:
 "Pyrex/ctypes/SWIG for wrapping image analysis"
 
It is best to let Damian subscribe to the list and give you the blurb directly.
 
Rgds,
Stephen> Date: Sat, 17 Jan 2009 21:59:43 -0800> Subject: Re: Speaker for Feb/Mar/Apr [Pyrex/ctypes/SWIG for wrapping satellite image analysis for nuclear non-proliferation]> From: damian.eads at gmail.com
> To: spmcinerney at hotmail.com> > Stephen, please strike the satellite imagery and nuclear> nonproliferation parts. 
> > Thanks,> > Damian> > On Sat, Jan 17, 2009 at 9:19 PM, Stephen McInerney> <spmcinerney at hotmail.com> wrote:> > I spoke with Damian Eads and he offers to give us an expanded version> > of his SciPy 08 talk for Pyrex/ctypes/SWIG for satellite image analysis for> > nuclear non-proliferation [link below]> > Not sure if Nathan or anyone else wanted to do a newbie nugget on this> > same topic that nght - could be good, I would encourage Nathan to.> >> > Jim or someone - can you mail out the list of our scheduled meeting dates> > for> > Feb-Mar-Apr at the new Symantec Mtn View location?> >> > Damian is based in New Mexico so let's see what fits his schedule.> >> > Regards,> > Stephen> >> > ________________________________> > From: spmcinerney at hotmail.com> > To: baypiggies at python.org> > Date: Mon, 29 Dec 2008 20:20:01 -0800> > CC: damian.eads at gmail.com> > Subject: Re: [Baypiggies] Opinions on Pyrex vs. ctypes vs. SWIG and whatever> > else binds to dynamic/shared libs> >> >> > The "which wrapper to use?" design question was in fact one of the topics> > of Damian Eads' paper at this year's SciPy 2008:> > "First Time Experiences Using SciPy for Computer Vision Research"> > http://conference.scipy.org/static/wiki/eads-cv.pdf> >> > Not saying the paper was the definitive word on this subject, but> > I thought it was a good paper and an interesting and topical application> > area (satellite image analysis for nuclear non-proliferation) and a> > SciPy newbie case study.> >> > I suggested we should invite Damian to present at BayPIGgies sometime,> > any comments?> >> > In fact, many folks have experience on this question, it would make a great> > topic for a presentation by multiple users sharing their viewpoints for> > different> > domains (I haven't done this stuff so I can't comment myself;> > The consensus answer as far as I can see seems to depend largely> > on how robust and stable the wrapperized code is, how much type-safety> > you need, exception-handling, how much control or visibility you have over> > the underlying source etc...)> >> > Stephen> >> >> From: nar at hush.com> >> To: baypiggies at python.org> >> Date: Mon, 29 Dec 2008 12:23:43 -0800> >> Subject: [Baypiggies] Opinions on Pyrex vs. ctypes vs. SWIG and whatever> >> else binds to dynamic/shared libs> >>> >> Hi guys,> >>> >> Long time lurker, first time poster -- I've got a question that has> >> been bugging me for a while and I figured I'd roll it out here to see> >> what opinions people may have.> >>> >> I've done a bit of work with Python binding to DLLs on Windows and> >> shared libraries on Linux/OSX, but I've mostly taken the ctypes route> >> using GCC-XML to generate wrapper code for me. It would seem that most> >> cheese shop modules use Pyrex with hand coded wrappers or SWIG to half> >> auto-gen the wrappers and finish off with hand fix-ups.> >>> >> Since ctypes ships with Python I kind of assumed it would be the> >> 'default' way of performing this task, but most people seem to like> >> Pyrex and a minority of people seem to use SWIG but hate it.> >>> >> I suspect the question is similar to asking 'Which is better, vi or> >> emacs?', but I'd like to hear what other people are doing / have done> >> and why they prefer it?> >>> >> For those unfamiliar with ctypes / ctypeslib here's the process:> >>> >> 0) Install gccxml> >> 1) Install ctypeslib module if it's not on your system, this will put> >> h2xml.py and xml2py.py on your system> >> 2) h2xml.py -I/path/to/includes SomeIncludeFile.h -o SomeIncludeFile.xml> >> 3) xml2py.py SomeIncludeFile.xml -o NewPythonModule.py> >>> >> There's a bit of fiddling you can do with the h2xml stuff as far as> >> preprocessing, but that's roughly what it takes. You still need to> >> hand-fix the generated wrapper code to make it 'pythonic' if thats> >> your flavor, since it passes C-type variables in and out, but it still> >> seems like a better development flow than using Pyrex/SWIG, but there> >> may be factors I'm unaware of. (Which is why I'm asking here!)> >>> >> Hopefully this isn't too off topic for the list, I look forward to the> >> responses,> >>> >> Thanks!> >>> >> Nathan Ramella> >> nar <> @ <> hush.com> >> >> >> > ________________________________> > It's the same Hotmail(R). If by "same" you mean up to 70% faster. Get your> > account now.> >> > ________________________________> > Windows Live™: Keep your life in sync. Check it out.
_________________________________________________________________
Windows Live™ Hotmail®: Chat. Store. Share. Do more with mail. 
http://windowslive.com/howitworks?ocid=TXT_TAGLM_WL_t1_hm_justgotbetter_howitworks_012009
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/baypiggies/attachments/20090118/c0113408/attachment-0001.htm>