W3C home > Mailing lists > Public > www-tag@w3.org > December 2007

Re: New draft of Elaborated Infosets document (xmlFunctions-34)

From: <noah_mendelsohn@us.ibm.com>
Date: Thu, 6 Dec 2007 12:44:34 -0500
To: "Henry S. Thompson" <ht@inf.ed.ac.uk>
Cc: www-tag@w3.org
Message-ID: <OFA831E01A.DC84417D-ON852573A9.005FABF6-852573A9.006113AB@lotus.com>

Henry, 

Thank you for the hard work on this [1].  I think it represents real 
progress.  I have a few comments on this draft:

----------
*  My main concern is that the goals aren't stated clearly, the ones that 
are there are to some degree contradictory, and in any case I'm not sure 
I'm comfortable with them.  From section 1: 

"That is, aside from the obligations imposed by the XML (and XML 
Namespace) recommendations themselves, what, if anything, ought to be done 
with a document whose media type tells you it's an XML document, before 
any application-specific processing is attempted?" 

This seems too imperative in style.  First do this, then do that.  I would 
prefer to see the goals grounded in the need to make Web documents 
self-describing.  So, it's important that there be an agreed, standard 
interpretation of an XML documents, and we here try to set out what that 
is.  The goal is not a set of steps, it's a preferred interpretation of 
the document.  The steps may be a way of achieving the goal.

"Or, to put it another way, if an author takes responsibility for the 
information in an XML document, exactly what is s/he taking responsibility 
for?"

Closer, but that's not at all the same as the first quote. and it's still 
somewhat oblique I think.

----------

* From section 4:

"There are three different ways in which the process of elaboration can be 
avoided, so that the unelaborated infoset is preserved: opting out, 
implicit quotation and explicit quotation. Opting out is trivial: Nothing 
in the definition of elaborated infosets requires a specification or 
processor to use it. So, for example, the next edition of XSLT probably 
should not mandate the elaboration of stylesheets, since on balance the 
presense therein of e.g. an xi:include element is most likely to be 
specifying a literal result element, and should not be elaborated."

This highlights my concern about the goals.  I thought the whole point was 
to have an interpretation that applies to >all< XML document, so that 
tools can be written that work across all of them.  If particular 
languages can opt out of the rules, what are the goals again? (see point 1 
above)

----------

* From section 2:

"The default processing model question can be rephrased as "Is there an 
infoset other than the one produced by a conformant XML parser which can 
and should be defined?"

I'm confused.  I would have thought the answer to that was "trivially, 
yes!"  I would have thought the question would be:  of the many that can 
be defined and will be defined (e.g. results of an XInclude transform), is 
there one that should be blessed as conveying the preferred interpretation 
of the source document?  Again, the goals need clarifying.

----------

* Question:  if someone invents a new quoting element in a coupld of 
years, say <bi:betterInclude>, how does this get worked into the 
elaborated infoset, and how to people writing software learn the quoting 
rules?

----------

* Section 4.2:

"The elaboration of an element II with this attribute is defined to be an 
otherwise identical element eII with the attribute removed, and the 
special property that it short-circuits further applications of E in 
search of a fixed-point."

I can't prove it, but my intuition is that the removal of the quoting (or 
keeping it for that matter) could be problematic if there are scenarios in 
which multi-level quoting is required.  Again, it's just an intuition, and 
probably not a good one.


----------

Overall, I'm a bit nervous that in spite of a lot of hard work on this, we 
don't quite know what we're doing yet, and we haven't shown that there's a 
mapping that applies in all cases (including XSLT stylesheets), that 
extends will as new transforms are invented, etc.  So, I'm a bit nervous 
about the whole undertaking, but quite willing to plug ahead and see 
whether we can get something good out of this.

Noah

[1] http://www.w3.org/2001/tag/doc/elabInfoset-20071127/

--------------------------------------
Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
1-617-693-4036
--------------------------------------
Received on Thursday, 6 December 2007 17:43:29 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:47:51 GMT