- From: Kjetil Kjernsmo <kjetil@kjernsmo.net>
- Date: Sat, 24 Apr 2010 23:10:18 +0200
- To: Semantic Web <semantic-web@w3.org>
- Cc: dev@lists.perlrdf.org, Dan Brickley <danbri@danbri.org>
On Sunday 18. April 2010 21:48:39 Dan Brickley wrote: > Perhaps you could talk through some RDF Perl stuff in IRC and email > as a stopgap? > > For my part, I've entirely lost track of what's doable in Perl these > days (having drifted to Ruby/Python mostly). I'll add a bit to what Toby and Greg just said: As they have stated, there is rough consensus that RDF::Trine and RDF::Query are the low-level APIs that most want to work with. I initiated a discussion to re-open a higher-level API in RDF::Helper, which is older, and also contains quite a lot of low-level stuff. I think RDF::Helper is valuable, and that most developers would want to work on the level of RDF::Helper, but it is an open question how we should do this. Some may be aware that Perl5 didn't have a well-developed object system, even though you could do OOP with it. Within just a couple of years, a new best practice has been established, and the community has gone "all the way" and adopted the Metaobject Protocol. This provides us with a new and very interesting direction: The MOP makes very straightforward to create classes at runtime, so, it should be possible to mix methods written in advance with stuff generated at runtime, based on the presence of an RDF Class, and based on RDFS subclassing and "subpropertying" :-) I haven't thought this fully through, but I have noticed some papers in the literature that explores this (e.g. one on using the MOP with the Common Lisp Object System). In Perl, we have Class::OWL, which implements some of these ideas, but there is a lot of potential for very interesting work here, and ultimately, we would like a MooseX::OWL, which would use the currently established best practice. I think this would be a very interesting student project, if anyone here is so inclined. :-) > What are the main opensource pure Perl tools available? How compliant > are they? How much of SPARQL (in memory, in SQL, in other things; > SPARQL protocol; SPARQL 1.1?) is covered? As you may know, Greg is on the SPARQL WG, so he's tracking it closely. I'm not tracking it quite as closely anymore, as I'm not a part of the group anymore, but I hope to create RDF::Query::HTTPBindings RSN... > How do the APIs compare within Perl, and across to other > scripting languages? I don't know the other scripting languages, but I wouldn't expect someone coming from a Jena background to feel lost with RDF::Trine/RDF::Query, since I have done that myself. > Are the tools happily packaged in CPAN? Yeah, some of it has not yet come out of github, but most of the stuff is on the CPAN and what is there is pretty mature, I would say. > (and Debian etc, if that's needed?). Redland, and RDF::Redland is in Debian, but nothing else is. I think it is too much in flux to be very useful in Debian, but that we may start thinking about getting RDF::Trine and RDF::Query in there once squeeze is released. > Are they well tested, actively > maintained? It is easy to see by looking at the latest release date. :-) I have not done anything with RDF::Core for example, and it is not actively maintained, so I would not recommend it. > Are there things written in other languages but missing in > Perl, which might easily be translated? Personally, I'm not so concerned abotu things being pure Perl. I think it is useful to be pragmatic about it: If things are fast enough in Perl, use Perl, if not, drop down to C/C++ or something, and Perl has well-performing solutions for it in XS. I would really like to see a low-level integration to 4store, so that it becomes a native RDF::Trine::Store, if it makes sense (it may not, HTTP- level integration is much lower hanging fruit). Also, it would be nice to have a port of OWLAPI to Perl, and a XS bridge to FaCT++. These would be on my wishlist, as I'm not very likely to do those myself. > Sorry again to hear this got cancelled. Amazing the amount of trouble > a volcano can make... ;) Yeah... I would really like to pick it up again. Possibly scale up! :-) Cheers, Kjetil -- Kjetil Kjernsmo kjetil@kjernsmo.net http://www.kjetil.kjernsmo.net/
Received on Saturday, 24 April 2010 21:11:21 UTC