RE: [xri] Property, rel comparison

["Scott Cantor" <cantor.2@osu.edu>]

Eran Hammer-Lahav wrote on 2010-01-26:
> As for the open issue in Web Linking regarding how to compare extension
> relation types, I will reach out to Mark, but if you feel strongly about
it,
> comments on the IETF Apps Discuss list would be helpful. We can't use a
> different rule than what Web Linking uses because the whole point is to
> share the same set of values.

Well, I took a look again, this time reading more closely, and I think it's
a problem that isn't really fixable because he's putting two types of values
(strings and URIs) into one field. That's a mistake, and I would have argued
against using "short names" for well-defined links. But I suspect I won't
win that argument, so there isn't much I can say, unless you think I could
make a case that using short link names is a mistake. I think it is, but I
know why people want it.

If you have to do case-insensitive compares of the IANA-registered strings,
then you can't really do case-sensitive compares of the URIs. I assume
that's why your comment on the that list was to do them as case-insensitive
compares even when they're URIs.

I can't really argue for doing XML namespace style compares as a result, so
we're kind of stuck. Worst of both worlds...we seem to need special rules
for Subject and special rules for rel types, but not the same special rules.

-- Scott