Re: comedi python wrappers

David Schleef wrote:
> On Mon, Apr 14, 2003 at 10:37:49AM -0500, Frank Mori Hess wrote:
> 
>> ... there already are python and perl wrappers for comedilib ...
> 
> Those are are pretty bad shape.  I'd welcome any work to make them 
> better, especially proper swig wrapping.

I did the Perl wrapper; I'm incredulous that it's still being 
distributed.  It hasn't been updated in 5 years and there were/are 
showstopper bugs in there which (to my knowledge) no one has ever 
complained about.  My conclusion from this is that essentially no one is 
interested in a Perl wrapper for comedi.

I have a slightly updated version that works for our application, which 
is stuck with an old comedi version as I simply couldn't keep changing 
our app to keep up with changes in comedi.

I originally did not want to use SWIG, as it was fairly new and I wanted 
more control from the 'hands-on' approach of XS.  I also wanted to be 
able to produce a more Perl-ish API, rather than a straight C to perl 
wrapper.

I'd say that if SWIG can really handle the whole wrapper, with no human 
intervention, then that's the way to go, if for no other reason than it 
will automatically keep up with comedi{,lib}.  There's even something to 
be said for slavishly copying the C API, as that's what most people 
using it will be familiar with.

Also, Bryan mentioned needing a DAQ card for testing.  Doesn't comedi 
include a 'null' device that's software only, just for this purpose?  If 
not, it should.  It would be a huge benefit to be able to do sanity 
checks on a system with out DAQ hardware.

<Joe

Received on 2003-04-14Z19:26:38