Style and read-only [was: Which style first?]

David Durand dgd at cs.bu.edu
Tue Mar 11 15:46:45 GMT 1997


At 9:10 AM +0000 3/11/97, Henry S. Thompson wrote:
And Terry Allen wrote:
>>  My question is perhaps off-topic here on xml-dev, and I know everyone
>>  is busy preparing for WWW6, but I ask you all to reflect on it as
>>  an issue that needs resolution later on:  What do I do to associate
>>  a style sheet with a read-only document, e.g., one that resides on
>>  some other server than my own, or that has been digitally signed?
>>  (And assume that this document has a doctype declaration already.)

First, I want to observe that Terry's point is very important... So we
really need to address it. It cuts to the core of why stylesheet
information needs to be loosely bound to a document. While I think that
binding style information into documents at all is a short-sighted
practice, what is more important is the ability to bind _new_ style
information onto the document _later_. If you have that you can always
ignore old, useless, or unwanted styles that are packaged with a document.

>Create a stub document with the SAME DTD which has a single top-level
>element which replaces itself (using XML-LINK) with the document you
>care about.

This should not work, as linking should cause stylesheet replacement -- and
adding stylesheet semantics to links is worse.

>Or if you don't like links, like this
>
><!DOCTYPE yours PUBLIC "http://<that read-only document's DTD" [
><!ELEMENT yours - - (rod-top-level-element-name)>
><!ENTITY rod PUBLIC "http://<that read-only document's body">
>]>
><yours>
>&rod;
></yours>

This doesn't work when &rod; contains a <!DOCTYPE ...> -- which was exactly
Terry's point.

I think that CATALOG-based proposals may be the best way to accommodate
such needs. Everything proposed for the style PI could fit as easily into
the catalog, and be more general, and less-tightly bound to the document.

>and in either case associate the style sheet with your stub in
>whatever way we end up agreeing on.

The problem is that you may not be able to create such a stub.


Here's the (practical) stylesheet problem the really bothers me:

HTTP 1.0 uses single connections per resource, and even HTTP 1.1 sends
resources serially down the wire, although it can re-use the connection.
This means that it will be hard to do incremental display of XML documents
unless we can get the stylesheet coming down the wire _before_ the document
itself. This seems problematic on several counts.

Since HTTP 1.1 is meant to make multiple connections to the same serer
unnecessary, the easy fix is ruled out by good network citizenship.

This is another place where getting a CATALOG could tell you quickly what
resources need to be fetched, and would let you get them in the right
order. I know that we hope that many stylesheets will be cached at the
client, but we can't count on that, especially from what I think I remember
about cache coherence on the Web.

  -- David

  -- David

_________________________________________
David Durand              dgd at cs.bu.edu  \  david at dynamicDiagrams.com
Boston University Computer Science        \  Sr. Analyst
http://www.cs.bu.edu/students/grads/dgd/   \  Dynamic Diagrams
--------------------------------------------\  http://dynamicDiagrams.com/
MAPA: mapping for the WWW                    \__________________________



xml-dev: A list for W3C XML Developers
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/
To unsubscribe, send to majordomo at ic.ac.uk the following message;
unsubscribe xml-dev
List coordinator, Henry Rzepa (rzepa at ic.ac.uk)




More information about the Xml-dev mailing list