DOCBOOK: xml:base

[Paul Grosso <pgrosso at arbortext dot com>]

At 02:12 2002 11 20 -0800, Bob Stayton wrote:
>On Tue, Nov 19, 2002 at 05:34:10PM -0600, Paul Grosso wrote:
>> Having xml:base is
>> only useful if there are relative URIs within scope, so
>> it's important to answer just what values are supposed to 
>> be affected by xml:base.
>> 
>> So just what things in DocBook are affected by xml:base?
>> 
>> Does xml:base affect the value of the fileref attribute?
>> What about ulink's url attribute?  They are both just CDATA,
>> so how is an application supposed to know whether they should
>> be affected?  Yet users will certainly expect them to be.

I'm okay with adding xml:base, but I still need to explore just
what it's supposed to mean to whom.

>Yes, fileref and ulink's url attribute would be affected,
>if they are relative references.  

Don't we need to say that then?  I would think so, but your
next statement seems to imply you don't this so.

>It shouldn't matter how a reference is specified.
>An application doesn't know an attribute is an external file
>reference until it is asked to open it up. 

It seems you have some processing model in mind, but I'm not
understanding it.  By what do you mean application in this case?
Who is asking the application to open it up?  And whatever the
answer, how does this thing know a given string is an external 
file reference?

I think the way xml:base is defined to work (both by the XML Base
spec and the Infoset) is to affect URI references within the
document's infoset.  But the fileref and url attributes are only 
attributes within the infoset, not URI references.

>Any such lookup triggered from within the context of an xml:base
>value should take the containing xml:base into account.

I don't understand how this statement translates into infosets.

There seems like too much magic--or "do what I want"--going on here.

I don't see how xml:base can work unless we say how to apply it.
I think at least we need to say which DocBook attributes are
considered to be URI references for the purposes of xml:base
processing.

>If the application is generating XML output, it should pass
>the xml:base value through so that downstream processors
>such as an FO processor can use it when it opens the file.

What application here?  Are you talking about an XSLT processor
processing an XSL stylesheet producing an FO result tree?  

In what way should xml:base be passed?  As an attribute?  
Attributes in an XSLT result tree are actively placed into
the result tree under the direction of the XSL stylesheet.
Nothing special about xml:base.

By the way, I also notice that the XML Base spec says:

  The deployment of XML Base is through normative reference
  by new specifications, for example XLink and the XML Infoset.
  Applications and specifications built upon these new technologies
  will natively support XML Base. The behavior of xml:base
  attributes in applications based on specifications that do not
  have direct or indirect normative reference to XML Base is undefined.

Given that the XSL spec doesn't say anything about XML Base
(and it is not based on the Infoset), the behavior of xml:base
attributes in XSL is undefined.

paul