Re: RDF APIs and Python

FWIW (and I'm coming in on this halfway), a uniform, low-level API  
for accessing RDF in Python would be a Very Good Thing, and would  
promote more innovation; e.g., things like sparta. Having CWM  
available for any store would be icing on the cake, indeed.

I guess the question is how to get from here to there; is rdflib  
acceptable to everyone? Should there be a PEP (seems kind of  
heavyweight at this stage)? Etc.

Cheers,


On 20/09/2005, at 7:06 AM, Chimezie Ogbuji wrote:

> P.S.  I'm forwarding this message to parties I think might be
> interested in such an effort.  So, please disregard if you aren't and
> I appologize for any overlap.
>
> On 9/20/05, Dave Beckett <dave.beckett@bristol.ac.uk> wrote:
>
>> I'd just like to say I am interested in common python APIs for RDF  
>> and
>> I've even licensed Redland's RDF.py more openly to encourage that, at
>> Daniel's request several years ago.
>>
>
> Excellent.
>
>
>>
>> However from my point of view, Python is just one of many important
>> language bindings for Redland, so I won't be going so far away  
>> from the
>> core Redland API scope, or have a lot of time to maintain one  
>> particular
>> binding over another.  Especially at present.
>>
>
> Understood.  The primary motivation for emphasizing Python was mainly
> because it is the low hanging fruit in making demonstratable progress.
>  Hopefully, such common/shared APIs can be written to be
> *implementation* and language agnostic.
>
>
>> The two posibilities that seem sensible to me in this area with  
>> respect
>> to Redland, are an adaptor class below Redland's RDF.py to call some
>> common API or an adaptor on top of Redland to allow a common API  
>> to have
>> Redland backend as an implementation.  (I think there might be  
>> some work
>> making Redland an rdflib backend but I can't remember the details)
>>
>
> Yes, there is (I used it as reference for doing the same for 4Suite as
> an rdflib backend):
> http://svn.rdflib.net/trunk/rdflib/backends/Redland.py
>
> I think having the common API above the implementation and an adaptor
> in between might be the preferred way to go (at least on the outset)
> since doing so requires that such an API is well-architected and
> properly abstracts common functionality across implementations.
>
>
>>
>> So this is more of a for your information message, and in a few  
>> months
>> time I might have the brainspace to see what the state is.
>>
>
> Thanks for the input.  If you do happen to have the brainspace anytime
> soon, I can simply add you as a user to the app server hosting the
> current (very basic) Markdown 'Wiki'
> (http://metacognition.info/markdown/RDFInterfaces.txt) so you could
> contribute directly.  Unless there is a better suggestion for a
> permanent home to facilitate this effort?
>
>
>> Feel free to forward or copy this message to any other RDF+python
>> interested parties.  Maybe also Mark Nottingham/sparta, ...?
>>
>
> Chimezie Ogbuji
> http://metacognition.info
>
>

Received on Wednesday, 21 September 2005 00:35:52 UTC