Re: [xsl] [XPath] matching elements which satisfy a specific content model

[Tobias Reif <tobiasreif@xxxxxxxxxxxxx>]

David Carlisle wrote:


> It would make some things shorter (but I still don't think that the kind
> of test in your exampleis often needed, your expanded example didn't
> show a need for it as it didn't show what you wanted in the other case)


When there is no title? Then I want no heading element; I'm testing it 
with xsl:if.

But again: In general, I think it should be possible to express simple 
content models with simple XPaths. You don't have to agree.


> However I think that actually this is making things conceptually twice
> as complicated. XPath is a language for specifying certain types of
> paths through an XML tree. Some simple cases are very simple and others
> are more verbose.


Not too verbose for you (if I understood you correctly), but too verbose 
for me :)


> DTD content model syntax is similarly a language for specifying certain
> kinds of paths through trees. The set of things that are simple and/or
> verbose in that syntax is unsurprisingly different.


It didn't say the difference would surprise; I just said that simple 
things should be simple in XPath, and that this *could* be inspired by 
DTD content model syntax.


> So it's not surprising that there are some things that are easier to say
> in DTD content models than in Xpath,


Sure it's not surprising.

> but I don't think the solution is
> to just have a combined language,


Not a combined language.

Just one additional function in XPath. Think about it without thinking 
of DTDs :)

> that would again make some things look
> a bit simpler but lead to even more complexity and lots of extra rules
> to tie down the semantics.


We choose XSLT over C in many cases :)

It could be specified with one line of EBNF, and one tiny para of text:
~
sequence: foo, bar
group: ()
one or more: +
zero or more: *
zero or one: ?
...

> In a dtd foo means the element with name
> "foo" in Xpath foo means the element with name foo in no namespace.
> What would "foo" mean in your proposal?


Since it's a proposed XPath function, why would it mean something 
different from what it means in XPath?

The DTD syntax example was to describe the actual content model, and to 
show how terse, elegant, and short it can be expressed.

Sure in my proposal foo means the exact same thing as it does anywhere 
in XPath; nothing to do with DTDs, which as you know don't know namespaces.

> Either you end up with something
> that looks like DTD syntax but has very different namespace conventions,


That would be no problem at all, since they would be the same as with 
all other XPath functions.


> or you have to introduce a whole new set of namespace unaware element
> name tokens into the Xpath language.


why?

> Either could be specified easily
> enough but it isn't really a simplification.


It is a potential simplification for the XSLT author.

Not for you maybe, but for me; A huge simplification in fact. You don't 
have to adopt my view, but you could respect it, as I respect yours. I 
don't want to tell you that you need content-model(); but I want to tell 
you that I'd be happy to be able to use it, without being told that "it 
isn't really a simplification".

The syntax could look familiar to those who write DTDs, but hat's the 
only thing my proposal has to do with DTDs. I do not expept the function 
to accept arbitrary DTD snippets.

Tobi


--

Vim users               donate.
http://iccf-holland.org/donate.html

Web developers           check.
http://www.pinkjuice.com/check/


XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list