[
Lists Home |
Date Index |
Thread Index
]
At 02:38 PM 12/26/2001 -0800, Soumitra Sengupta wrote:
>I agree with both your points about strong typing and functions. Where I
>disagree is in the priorities. A significant number of business
>applications are
>being written using DTD's where the notion of typing does not exist.
True - but a significant number of business XML applications are being
written in a hodge-podge of languages with differing type systems, and in
some environments, this is getting to be a bit of mess. Companies who
maintain mission-critical company data tend to use a database to protect
the integrity of their data. Few companies would like to keep important
data in a system that could overwrite an invoice with a goose and not
notice. That's why type safety matters - and especially if update, insert,
and delete are allowed.
>Functions
>can and should be dealt after UPDATE. What is the point of INSERT ing
>something
>when I do not have a standard mechanism for altering it or UPDATING it. I
>think
>this limits use to the "read-only" world as far as applications are concerned.
Perhaps I was unclear - by update, I mean insert, update, and delete. I
think we should either do all of these if we do any update proposal at all.
>At the beginning of this discussion, I was not sure where I stood. Now I am
>certainly in the "some UPDATE semantics even if incomplete" before XQuery
>gets to
>recommendation camp.
The first step is to generate a solid proposal. I'm involved in generating
one right now, together with several other companies. Once we have the
proposal, the Working Group will have to decide how solid it is, whether it
has time to review it, etc.
During a town hall meeting at XML 2001, four panelists were asked when they
thought we should try to release XQuery, and most answered that we should
try to release it this summer. I would love to see us have updates in
XQuery 1.0, but I am very skeptical that we can add them and be done in
this time frame. If we need, say, six months to add updates, I would rather
see an interim release six months after XQuery 1.0.
Of course, if we find that we do have time, I would love to slip it into
XQuery 1.0. But to me, an underlying type system is important to allow
people to make updates in a consistent and safe manner, so I want the type
system first. And I think we are in the home stretch on this work.
As for functions, we have had functions since the beginning, and the only
real remaining issues with functions are the type system issues. We need to
solve these anyway, and functions tend to help focus attention on the
issues that need to be resolved.
Jonathan
|