[XML-SIG] RDF Parser -> PyXML

[Uche Ogbuji]

> Martin v. Loewis writes:
>  > 1. PyXML user who also uses RDF. I'd certainly appreciate if I had to
>  >    install only a single package, not two - unless the release
>  >    schedule for PyXML would compromise timely delivery of RDF code
>  >    updates.
> 
>   Which indicates that the RDF code should be separate, at least until
> it has stabilized and has good reception from users.  I don't know
> where Redfoot is on the development curve.

Hmm.  I don't know that we can claim much history of waiting until code is 
mature before we merge it in.

>  > 3. PyXML maintainer and packager. I appreciate contributions from
>  >    whoever is willing to contribute, as long as it is open source, and
>  >    as long there is willingness to maintain the contribution for the
>  >    months (and years) to come. My biggest concern is that the original
>  >    author runs away and leaves me with all the bug reports.
> 
>   Reinforces my comment on #1 -- don't include until it's stable.
>   Perhaps the right thing to do would be to have a separate package,
> with an option to consider merging it in later.  I think it would be
> good to keep development discussion here -- it *is* a general XML
> thing, whether or not all of us use it (we don't all use validation
> either, but it's in the package).  The cross-pollination and expansion
> of the set of use cases is good.  I don't mind sharing the CVS
> repository, either -- CVS supports multiple modules quite well.
>   Why don't we set Redfoot up as the "redfoot" CVS module and reserve
> the package name xml.rdf for it?  (As in, let it require PyXML and
> then install itself to ../site-package/_xmlplus/rdf/.)  If a decision
> is made to keep it separate or merge it in later, there are no real
> disruptions for user code.  Is this reasonable, or am I off my rocker
> again?

Well this is all fair enough, but the only problem is that there is little 
value to it if it doesn't get packaged in.  If it doesn't, there aren't people 
working with it, finding bugs, fixing them, etc., and it won't magically 
become mature sitting in a dungeon.

I think it might help this discussion to note that James's code is really just 
a SAX handler that emits RDF triples in simple Python form.  It's by no means 
a monster.  And it's by no means a full-blown RDF system.  4RDF consists of 
almost 100 Python files, only *one* of which implements a parser.  Checking in 
Redfoot's parser would provide a simple, common root from which the (hopefully 
many) Python RDF implementations can flourish, packaged separately.  Sort of 
like an expat for which many different DOMs have grown.

Of course, my opinion is to go ahead and check it right into PyXML as an 
"xml.rdf" package.  After all, it only deals with the XML serialization of RDF.


-- 
Uche Ogbuji                               Principal Consultant
uche.ogbuji@fourthought.com               +1 303 583 9900 x 101
Fourthought, Inc.                         http://Fourthought.com 
4735 East Walnut St, Ste. C, Boulder, CO 80301-2537, USA
Software-engineering, knowledge-management, XML, CORBA, Linux, Python