Re: is "foo element(foo/1/3)" a valid xpointer?

At 23:45 2003 04 15 +0200, Daniel Veillard wrote:
>On Tue, Apr 15, 2003 at 04:21:11PM -0500, Paul Grosso wrote:
>
>> In particular, suppose we want to point to element(foo/1/3)
>> but fall back to just pointing to the element with id=foo
>> if the element scheme isn't supported.  I'd expect to write
>> something like:
>> 
>>       href="mydoc.xml#foo element(foo/1/3)"
>
>  Hum, that's the way around, aren't schemes evaluated from
>left to right ? That would not work anyway.

Ah, you're right about the order.  So what I want to work is:

  href="mydoc.xml#element(foo/1/3)foo"

>> But reading the BNF at [1], it looks to me like Pointer
>> can be either a Shorthand or SchemeBased, but not both,
>
>  that's my understanding too.
>
>> and SchemeBased consists of PointerParts that each
>> require a SchemeName, so I don't see how what I show
>> above can be allowed by this grammar.
>
>  I don't think you can't expect any fallback mechanism 
>with the current set of specs if element() is not supported.

Well, it's not just if it isn't supported.  It's also if
the given element child sequence has a resource error
(that is, fails to find an element).  And I really think
we want to be able to fall back from element() to shorthand.

Does anyone else remember if we did this on purpose or if
we meant to allow falling back to a shorthand pointer?

paul

Received on Tuesday, 15 April 2003 17:57:06 UTC