Hey Philip. On Nov 22, 2007 12:49 AM, Philip Cooper <[EMAIL PROTECTED]> wrote: > I have made some upgrades to RDFAlchemy. RE:
Excellent, I had earmarked it as an alternative to Sparta for more "pythonic-RDF" APIs but wasn't sure if the work there was still active. > http://www.openvest.com/trac/wiki/RDFAlchemy > At it's core RDFAlchemy is an ORM layer on top of rdflib. > I have added (early buggy) support for read/write of Sesame2 and read > of SPARQL Endpoints. Nice :) > In the process of doing that I did so by using the rdflib api and > writing the appropriate methods...mostly just triples and query (and > add for sesame). Might be of interest to folks here and as I stand > back and look at it, SesameGraph is almost a drop-in replacement for > ConjunctiveGraph (not nearly as mature obviously). I'm not sure about: It would be nice if we had some semi-formal writeup of a Python "style" or API for the RDF / N3 abstract syntax. It doesn't really matter who writes it or what it is based on. I mainly use list comprehension on graphs, infix operators, triples, query, and some of the other agile ones: subjects, predicates, objects, triples_choices. But that's basically it for me. I"ve never really felt that (with Python) the differences between existing RDF APIs (cwm,euler,rdflib,Redland python bindings, *and* now trio) is significant when you don't consider the really idiomatic APIs (such as generators, iterators, etc..). It would be nice if we had a PEP for RDF APIs or something similar. > * should I be using rdflib as a pseudo standard api? As someone who has been recently trying to build more abstract APIs on top of rdflib (InfixOWL is a Pythonic OWL API and FuXi is a Logic Programming Python API), I've found the rdflib 'base' a useful archetypal API (this is why I switched over from 4Suite RDF a looong time ago.). In the interim, I'd say suggest "yes", but it would be nice if we had some well-written, idiomatic RDF APIs for Python that tool-writers can adhere to universally. It would be a shame if we were never able to better leverage the host language's expressiveness in common ways > * if this works...should it be rolled into rdflib? +1 (if we continue with the current package management) -1 (if we go with earlier suggestions to break rdflib SVN into smaller chunks) > Plenty of things in it that are probably not rdflib-ish enough (I made > up my own mind about query responses since I don't use them at all in > my rdflib coding). It has been a while since I read RDFAlchemy documentation, I'll take a closer look again now that you have updated it. > Thanks for the great work on rdflib. Just trying to push the ball > forward. Likewise. _______________________________________________ Dev mailing list Dev@rdflib.net http://rdflib.net/mailman/listinfo/dev
