xml-dev - RE: Why the Infoset?

[Aaron Skonnard <aarons@develop.com>]

> -----Original Message-----
> From: Simon St.Laurent [mailto:simonstl@simonstl.com]
> Sent: Tuesday, August 01, 2000 2:36 PM
> To: xml-dev@xml.org
> Subject: Re: Why the Infoset?
>
> I'd suggest that the 'whole idea of the Infoset' is effectively throwing
> out the baby with the bathwater, by discarding far too many details.
>

It seems to me that most of this debate is due to the fact that the Infoset
was developed *after* XML 1.0 + Namespaces. Had the order been reversed, I'm
not sure things would have turned out different but the critics may have
found it more acceptable.

While teaching XML courses, I've seen a number of companies using the
benefits of the Infoset to achieve higher-levels of interop, without the
data ever looking like XML 1.0 (e.g., the output of one system is a stream
of SAX-based calls that are rehydrated as a DOM tree on another system and
navigated using XPath, transformed using XSLT, etc.). As long as everyone
understands the same data model, we can achieve better interop - bottom
line.

I'm not convinced that XML 1.0 + Namespaces will forever be the canonical
serialization format. It may not change any time soon but if this
possibility exists, I would rather see the Infoset err towards the abstract.
Nothing in Appendix C bothers me - placing to many serialization details in
the Infoset restricts future possibilities.

-aaron