This is the mail archive of the
docbook-apps@lists.oasis-open.org
mailing list .
Re: Re: needing clarification about XSL transformation
- From: "Robert P. J. Day" <rpjday at mindspring dot com>
- To: Markus Spath <mspath at arcor dot de>
- Cc: docbook-apps at lists dot oasis-open dot org
- Date: Mon, 03 Mar 2003 10:35:42 -0500 (EST)
- Subject: Re: DOCBOOK-APPS: Re: needing clarification about XSL transformation
On Mon, 3 Mar 2003, Markus Spath wrote:
> Robert P. J. Day wrote:
> > On Mon, 3 Mar 2003, Vitaly Ostanin wrote:
> >
> >
> >>With this style xslt-processor must not copy comments and PI.
> >>This style not overriding built-in templates, so saxon is
> >>incorrect.
> >
> >
> > ah, so as i read this, the conflict resolution is that,
> > even if i have a template that matches "node()", that will
> > be overridden by the more explicit built-in rule that matches
> > "comment()" explicitly, whose effect is to do nothing with
> > the comment.
> >
> > perhaps it's just kay's wording, but in his book at the
> > bottom of p. 315, he writes (after a list of how template
> > matching is done):
> >
> > "If there are *no* [my emphasis] templates that match
> > the selected node, the built-in template for the relevant
> > node type is used."
> >
> > the way i read this is that the "node()" test *would*
> > match a comment(), and thus my template would be used.
> > apparently, that's not what he meant, but you can see
> > how it could be interpreted that way, i hope.
>
> but that actually would imply that kay contrdicts himself.
>
> XPath rec says: A node test node() is true for any node of any type
> *whatsoever*
i definitely don't buy *that*, since most of the docs are adamant
that, at the very least, an attribute node would not match node()
since, technically, an attribute node is not considered a child
node. i'm getting confused again.
>
> anyway: if you are really interested in resolving this issue, i'd suggest you
> post it at the xsl mailing list, both kay and veillard (author of xsltproc, if
> i remember right) will probably be eager to proove their transformer is conformant.
i think i'll do just that.
> or you take a pragmatic approach and just include comment() and pi(),
> since the w3c recommendations are ambiguous sometimes.
i'm doing that now, but it sure would be nice to nail this
once and for all.
rday