Re: issue with relative URIs in the DOM

I guess not sending the message to the www-dom mailing list was intentional,
but strictly because my first message didn't appear into the archives and
thought I wasn't getting through.  Not sending it to the XML Linking
comments list was an oversight.  You were right, when I said parameter
entities and meant external entities.

It does appear that all the underlying needs are on the DOM issues list,
however it would still be good to mention in XML Base's "Impact on other
Standards" section.

Philosophically, I think the scoping statement betrays the objective of XML
Base.  XML Base's motivation was to separate resolution of relative URI's
from the physical location of the resource.  But the scoping rule, makes the
physical structure of the XML document an issue in interpretation.

As Joe outlined in a previous message, the concept of a document's (or
entities) base URI definitely becomes problematic when you are dealing with
a transient document in an applications' DOM.

The DOM isn't the only technology that feels free to lose the entity
boundary.  If you ran a document with external entities through an XSLT
processor with even a simple copy transform, the resulting document would
have lost all the entity boundaries.

Since the entity boundary is such an easy piece of information to lose (or
better, such a hard piece of information to keep), it seems that the scoping
rule is just too hard to maintain through multiple steps of processing.



----- Original Message -----
From: "Paul Grosso" <pgrosso@arbortext.com>
To: "Curt Arnold" <carnold@houston.rr.com>; <keshlam@us.ibm.com>
Sent: Friday, July 14, 2000 9:44 AM
Subject: Re: issue with relative URIs in the DOM


> At 21:38 2000 07 13 -0500, Curt Arnold wrote:
> >I did a quick reread of the of the EntityReference section in the Level 2
> >candidate recommendation.  The prose doesn't indicate that there is any
> >standard mechanism to control whether a parser represents an entity
> >reference with an explicit EntityReference node or expands the reference.
> >It seems that the behavior is totally at the discretion of the of the
parser
> >implementation.
> >
> >Then the XML Base document has this: "The scope of xml:base does not
extend
> >into external entities, but it does extend into internal entities."
> >
> >It appears that it is not possible to write code that would work
implement
> >XML Base reliabily with any conformant DOM implementation.  You could
write
> >code that would handle the scoping, but it would only work with DOM
> >implementations that did not expand the entity references so that you
could
> >get to the system identifier.
> >
> >It does to seem to be an issue that should be at least mentioned in XML
> >Base's "Impacts on Other Standards" section.
>
> I think this is reasonable, but I suspect the XML Base editors would
> need help from Joe and/or others more familiar with the DOM than I to
> know how to word it.  Note that the problem isn't with XML Base per se,
> as Joe discusses, it's also a problem in the absence of XML Base in that
> the DOM does not currently surface all the base URI info needed to resolve
> relative URIs.
>
> >The options seem to be:
> >
> >a) acknowledge that that there is an issue with implementing XML Base on
DOM
> >but not address it further in either standards.
>
> Acknowledgement seems a reasonable thing.
>
> >b) Add some functionality to DOM to determine absolute URI's from a
relative
> >URI.
>
> That is a DOM question, and Joe has commented.
>
> >c) Remove the scoping statement from the XML Base draft.
>
> This just wouldn't help.  Even without XML Base, an external entity
> has a different base URI than the document entity, and relative URIs
> within the external entity are relative to that different base URI.
> XML Base's scoping is merely inherited from existing base URI scoping.
>
> >d) Mention that xml:base scoping across parameter entity boundaries may
be
> >an issue and recommend that top level elements in parameter entities have
> >explicit xml:base attributes with absolute URI's.
>
> I don't think you mean parameter entities, but external parsed entities.
> And the xml:base scoping isn't the issue (per my previous paragraph).
> Use of an absolute URI in a top level xml:base would make it clear how
> to resolve any relative URIs within the entity, but that's often not
> what is wanted even now in the absence of XML Base.
>
> >e) Provide a mechanism to force an explicit entity reference node in the
> >DOM.
>
> That is a DOM issue.
>
> >Option a) seems to be minimum acceptible response.  Option b) is
> >sufficiently difficult that there is not an obvious quick fix.  d) would
> >seem to be a prudent hack.
> >
> >Option c) apparently has been a contentious issue (see
>
>http://lists.w3.org/Archives/Public/www-xml-linking-comments/2000JulSep/005
6
> >.html).
> >
> >I can't really debate the issues, just that I don't think you can
implement
> >XML Base on DOM Level 1 or 2 which would seem to be a significant issue.
>
> What Joe is saying is that you can't implement proper handling of
> relative URIs in external parsed entities even in the absence of
> XML Base with DOM Level 1 or 2 (right, Joe?).  The way I see it,
> it's just not an XML Base issue.
>
> But I will suggest that the XML Base editors add a mention of the
> DOM to the appropriate appendix of the XML Base spec.
>
> Do I have your permission to forward a copy of this message to
> the www-xml-linking-comments@w3.org list so that this request
> gets correctly archived?
>
> paul
>

Received on Friday, 14 July 2000 12:45:28 UTC