James Robertson wrote:
So what people 'believe' is irrelevant. Facts are stubborn things.
Now, whether that's the best mechanism for extensibility? That's a good question. I'd have to say that the simple "It's worked so far" argument carries a lot of weight.
That's a good point, though calling them "modules" is kind of overblown. A more interesting question might be which common extensions would be invalid RDF in RSS 1.0. Almost all extensions seem to consist of a single element with a text value. RSS 1.0 seems quite popular, so I wouldn't say the victory is decisive for either approach.
Well, that raises a question - I don't know the answer to it though. Do most aggregators that deal with RSS 1.0 really do RDF, or do they just treat it as a variant syndication format? I know that I didn't bother creating a distinct ability to deal with RDF in BottomFeeder - I'd be curious as to how others dealt with this one.
I think I know the answer to that question. Almost none of them treat RSS 1.0 as RDF. That doesn't mean that producers don't find it convenient, or that the syntactic restrictions of RSS 1.0 don't help interop.
All variants of RSS are essentially pseudo-RDF, and we've had a hard time saying anything enlightening about extensibility. I think taking another look at RDF might be worth it (look at the category thread), as long as we can keep the syntax under control.
There are certainly downsides to RDF when it comes to validation, so we would definitely benefit from an approach similar to the DOAP project[0], where the format must comply with an RNG schema (or XSD, Schematron, and whatever else Randy feels like writing :).
I think Henry's suggestion A is worth looking at, though the link element might have some problems. RSS2-style extensions would still be as straightforward as ever.
Robert Sayre
[0] http://www-106.ibm.com/developerworks/xml/library/x-osproj3/
