Re: SYMM WG comments on XBase 2nd last call

At 18:34 2000 07 07 +0200, Philipp Hoschka wrote:
>"Cohen, Aaron M" a écrit :
>
>> Philipp:
>> I didn't realize that you had to define what elements could take xbase.
>> Doesn't it work with general XML, that doesn't have a semantic spec or a
>> specific DTD? I've been looking at it as a sort of XML extension, somewhat
>> indepedent of the specific XML application language.
>
>i think you have to specify in the DTD where you can use xml:base; similar
>to xml:lang, and xmlns, actually - if i'm wrong, i'm sure the xml linking
>folks will correct me

I think that's right in that an XML 1.0 DTD requires all attributes to
be declared for the document to be "valid" (in the precise meaning of
"valid" in XML 1.0).  This is an XML 1.0 thing, not an XML Base thing.

One could use xml:base (at most once) on any element in a well-formed 
XML document.  (Also an XML 1.0 thing.)

As far as an application knowing where to *use* xml:base, it is up
to the application (e.g., SMIL) to describe how to recognize what bits
of a document are (potentially) relative URI references to which xml:base
should apply.

>> But if we do need to specify, we can certainly just say that it's valid
>> (although not necessarily meaningful) on all SML elements.
>
>i guess so - i would hope that it is meaningful, though  it may not be
useful,

Right, your DTD should probably allow xml:base on all elements, and 
it is meaningful (as defined by the XML Base spec) everywhere:  its
meaning is to define the base URI within its scope.

As far as where the base URI is used, it's used to help resolve any 
relative URI reference (per RFC 2396), but it's up to the application 
(e.g., SMIL) to explain how to determine what strings in the input 
document are to be interpreted as relative URI refs.

>but that's another issue - quickly checking the draft, it seems that xml:base
>*is* actually meaningful wherever you chose to put it

Yep.

paul

>> -Aaron

Received on Friday, 7 July 2000 13:16:39 UTC