[
Lists Home |
Date Index |
Thread Index
]
Fair enough. I don't really smell conspiracy here. I hope that it will not
be technically or politically impossible for vendors to use RELAX NG
provided that RELAX NG does the right thing wrt type assignment.
Mike
Rick Jelliffe wrote:
> XQuery was under way (in some form) before RELAX was developed, and
> certainly before RELAX NG. Just as RELAX came too too late to
> influence the fundamental stages of WXS, so RELAX NG is too
> late to influence the fundamental stages of XQuery.
>
> Furthermore, the point of XQuery (rightly or wrongly: depends on your use
> cases and your vision of the Web) is to make use of types: RELAX
> NG has been
> explicitly not about types in that same way. So there is a
> technical impediment
> too.
>
> But now RELAX NG is out, and clearly viable and attractive, there
> are different
> dynamics. First, the RELAX NG people are, I think, looking at
> various ways
> that RELAX NG could be more useful for queries. Second, there is
> discussion
> at the W3C to disconnect XQuery from WXS validation (see Tim Bray's
> comments on the TAG public list). Such a Type-augmented Infoset
> (TAI rather
> than a PSVI) would still be based on the categories established in WXS,
> but it would be clearer that the augmentations could come from different
> schema languages. I expect that this is the way things will with XQuery
> and subsequent revisions of WXS.
>
> So we shouldn't see XQuery per se as a conspiracy against RELAX NG.
> However, the general idea that an XPath 2 (in particular) implementation
> would require a WSX validator does unfairly privilege WXS (and a TAI),
> and that is the wrong direction.
>
> Cheers
> Rick Jelliffe
|
