See also: IRC log
No regrets heard
Norm: I like it. It's a little confusing but that's not our fault.
Vojtech: It seems to me that we
have six classes (A, B, B', P, V, and X) and then a list of
which information items exist and which classes they apply
... But B and P are always used together so why can't we just merge them?
... Also, if you look at the profiles, there are no conflicts.
Norm: I think A, B, and B' are about what the profiles provide; P, V, and X are about the items provided by the underlying processor. I think it would be a category error to combine them.
Vojtech: Oh, ok.
... If someone wanted to introduce a different profile, maybe they'd need the distinction.
Alex: And it's a clear indication
of where things could be simpler.
... If I had a particular processor and I wanted to see if I conformed, that would require merging states like V and P.
Norm: They things you have to provide to conform are enumerated in the sections above. For example, 2.2 says your underlying processor has to provide P and X.
Alex: None use V
Norm: No, we don't have a profile
that requires DTD validation, which I think is the right
... Paul, I know you asked some folks from XML Core to review it carefully, but I don't think that's happened yet.
Paul: That's right.
Vojtech: I asked about the references property, but I guess that's still unresolved.
Norm: No, under Attribute Information Item, I think that's resolved.
Vojtech: Oh, ok, I see. Yes.
Norm: Any other discussion?
I propose that we give the Core folks a week or so to review. If something significant comes up, we'll hold off on publication. Otherwise, we republish this as a new Last Call WD sometime early in March.
Norm: Any objections?