Castor (for Java, from www.exolab.com), uses an actual XML Schema for this. The advantage is that you can leverage off the fairly rich existing set of defined datatypes. It would be nice to see "xml marshalling" (as they call it) integrated into an existing Perl object-relational framework like Tangram. OTOH, Tangram isn't very well optimized for mod_perl work in general--currently the object cache must be completely flushed after each request, etc. I think an open-source mod_perl-oriented O/R system with XML support would be extremely useful. I would be interesting in hearing opinions on whether extending Tangram and its ilk is viable. Either way, I will probably look into working on this in the near future, let me know if you're interested. --Chris On Tue, 30 Jan 2001, Matt Sergeant wrote: > On Tue, 30 Jan 2001, Robin Berjon wrote: > > > >Well OK then, lets look at a way to dump current objects to XML, or > > >interface to what you've got. I think that could be useful to a lot of > > >people (myself included). > > > > Yes that would be indeed very interesting. It can't be totally automagic > > *and* be useful in the context that I (and I guess others) need, which is > > to dump an object into a vocabulary that makes sense wrt the app. The way I > > see it is that it's more or less schemata the other way round: given an > > infoset (more or less) and stuff to fill it with, and produce a document. > > Maybe a small language defining that would be both possible and good. > > In fact it could be an XML vocabulary, similar to TREX, something like: > > <px:perl_to_xml xmlns:px="urn:to-be-decided"> > <employees> > <px:array> > <employee> > <px:hash key="name"> > <name><px:scalar/></name> > </px:hash> > <px:hash key="department"> > <department><px:scalar/></department> > </px:hash> > </employee> > </px:array> > </employees> > </px:perl_to_xml>
