xml-dev - Re: Why the Infoset?

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

At 02:37 PM 8/2/00 -0400, Paul W. Abrahams wrote:
>The debate over the Infoset might be clarified and sharpened by
considering the
>following normative statement, to be added to the Infoset spec:
>
>Whenever two XML textual constructs map into a single construct of the
Infoset, an
>XML processor must produce identical output for those textual constructs.
>
>Personally I favor this, but not with any solid sense of conviction.
Simon St.
>Laurent in particular has advanced some cogent arguments against it.

I think we've just encountered a broad plain of possibilities with few
natural landmarks, and establishing where to put the boundaries is difficult.

I can't get excited about representing all 74 instances of whitespace in
the XML specification.  I can get very excited about providing information
from DTDs that might be useful to applications.  (See 203-5 of my _XML
Elements of Style_ for some detailed reasons why, but we can just say I
have doubts about self-describing documents.)

I don't find the proposal you put forward very much different from the
existing Infoset, though it makes stronger claims about conformance.  

The 'parsed results' school of thought, I'm afraid, is not very interesting
to me.  Parsers already do too much and lose too much along the way for me
to like that boundary.  

I've addressed the impact of this a few times already:
http://www.simonstl.com/articles/layering/layered.htm
http://www.simonstl.com/articles/interop/
http://www.simonstl.com/projects/xpdl/xpdl2/index.htm

It doesn't seem to be getting through to the decision-makers at the W3C
very well, though it's gone over well in a number of other contexts.

Oh well.  Next thing I'll be going on about why groves are important...


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