Tim Bray wrote:
On Jan 8, 2005, at 8:23 AM, Bill de hÓra wrote:
My answer to this question is that Atom doesn't have a model in terms of being able to talk about extension so there's no point discussing it. Extensibility is probably out of scope for the format.I'm not going to let that go unchallenged. The ability to put your own markup in a feed entry and be confident that it will not cause receiving software to break, and that it will be there for receiving software that actually knows how to deal with it, is what many people describe using the word "extensibility". I think that Atom is, in a popular usage of the term (a large majority usage, I'd bet, but without backing statistics) extensible.
The ability to add new structures that have not a lot to do with the existing structures sounds like modularity. Being able to process what is expected to be there in the face of stuff that is not expected to be there sounds a lot like robustness.
We could go at this all month. In truth, I have no problem agreeing to your use of the term.
The people who insist that you have to have to buy into a graph-theoretical KR model to be able to use the word "extensible" live in a universe that is self-consistent, but it's not the one where I work.
Look, the point is this. Those arguing from the RDF side of the house do mean what you mean by extensible. Furthermore, what is meant there by extensible hasn't been demonstrated (in my mind) as a requirement for Atom. Thus, in some respects we're having a pointless discussion. One side or the other is going to have to give up possession of the term extensible, or we're going to continue to watch people talk past each other. Once you appreciate that we are talking about two different things, execution of the work in hand will become simpler.
cheers Bill
