xml-dev - Arbitrary Infoset boundaries (was Re: Common XML - Final Review Draft)

["Simon St.Laurent" <simonstl@simonstl.com>]

At 05:00 PM 8/2/00 -0400, John Cowan wrote:
>> I have no objects to subsetting XML.  I do have objections to subsetting
>> the abstract representation of XML without subsetting XML itself.
>
>Simon, you can't have it both ways.

I'd suggest that the W3C is trying to have it both ways - they won't subset
XML, but heck, they'll subset the Infoset instead!

I suppose this will make defining an Infoset for Common XML core very very
easy, but I'm not sure it's good for the rest of the XML community.

>Either the Infoset is a subset of the full information available in the
>text of an XML document, or it isn't. If it is a subset, then someone has
>to choose the subset.  If it isn't, then it has to provide the full
information
>for reconstruction, including the 74 different whitespace properties.  Which
>you say you don't care about (and rightly so IMHO).

I'd suggest that's pretty arbitrary, and that your use of '74 whitespace
properties' is basically a barricade to keep out any of the rest of the
items in Appendix C.

>One sticking point seems to be DTDs.  If we did DTDs, we'd have to do
>Schemas too, and that seemed like too big a job (plus the Schema WG is
>changing things all the time as well).  So we arbitrarily ruled
metainformation
>(except for base URIs) out of scope.

You arbitrarily ruled it out of scope, the DOM folks arbitrarily ruled it
out of scope, and now we have a lot of XML features that applications just
plain can't get at easily.  

And how exactly did base URIs creep back into scope, while DTD dropped?  I
can't find 'base URI' anywhere in the XML 1.0 spec.  Just curious on that
one.  (Schemas aren't in either XML 1.0 or namespaces, so I'm much less
inclined to dispute that boundary.)

If you want to represent XML 1.0, represent it.  If you want to represent
Common XML, represent that instead.

>> Is the Infoset about XML 1.0+Namespaces, or is it about something else?
>
>It's about what's meaningful in XML 1.0+Namespaces, as best we can figure it.

What I'm saying here is that 'best we can figure it' appears to be a purely
political decision that has some significant consequences.

I'd suggest that it's better to drop that decision making - and drop the
Infoset entirely - than to continue the current approach.  Sorry for all
the blood you've sweated, but that's what I feel quite strongly.  

I seem to remember telling you all this on bench in Washington Square Park
over a year ago, but maybe I should have taken it public then.

Simon St.Laurent
XML Elements of Style / XML: A Primer, 2nd Ed.
http://www.simonstl.com - XML essays and books