Re: [xml-dev] IDs considered harmful or why keys might be betterthanIDs...

[Sean McGrath <sean.mcgrath@propylon.com>]

[Jonathon Borden]
>In our rush to 'move beyond' DTDs we sometimes neglect
>the lowly internal subset which _is_ instance data, so for example
>specifications like SOAP which are now stuck with the inability to have IDs
>(shrug).

The case against internal subsets is compelling IMHO. They don't round
trip. This creates significant problems for software developers
working with hetrogenous, loosely coupled XML processing
systems. i.e. anything other than XML "viewers".

I have droned on about this for years now and
will spare everyone a replay[1] :-)

I believe (correct me if I'm wrong James) that James Clark
is referring to this problem when he says of the internal
subset approach to this problem ([2])

"(iii) It doesn't work with streaming output. This is in my view the most 
important technical problem"

regards,
Sean


[1]http://lists.xml.org/archives/xml-dev/200109/msg00722.html
[2] http://lists.xml.org/archives/xml-dev/200111/msg00461.html