RE: nasty namespace issue (was RE: Is minimalism a goal?)

>
>
> > What exactly is "wrong" about MS's XSL interface?  Would "Joe Webmaster"
> > care, or is this a more theoretical standards compliance issue?
>
> As Aaron said, the xpaths seem to use the default namespace.  Joe
> Webmaster will care if his XPaths work one way with the DOM, and another
way in XSLT
> processors.  This comes up all the time when using XPaths.
>
> -scott
>
>

Exactly. The problem with attempting to address this issue in the "ultra
minimal" API (that doesn't support some type of namespace context/bindings),
is that there is NO WAY to query documents that use default namespace
declarations (see the example in my last post). This must be the reason
MSXML did what they did. I just tested Oracle's implementation of
selectNodes and they do deal with namespaces properly (they don't consider
default namespace declarations), but they do provide an overloaded version
of the function that takes a second parameter to an NSResolver interface.
This is closer to what we need.

Hence, if the "ultra minimal" API doesn't address namespace bindings, there
is no way to do it properly without making it impossible to deal with
documents that use default namespace declarations. In other words, I think
the "ultra minimal" API MUST address the namespace issue.

-aaron

Received on Monday, 8 May 2000 10:36:14 UTC