RE: Future of XQuery and XQuery Update Faclilty

[Thomas Lord]

Jonathan Robie wrote:
>
> Perhaps another way of putting it: what is central here is the updates 
> that are to be done. To specify the updates to be done, an application 
> would write queries that contain updates, it would not generate a 
> pending update list in XML so that a different application can do the 
> updates. Distributing a single update among multiple applications is 
> not a simplification.


Ugh.   This is the problem of people not wanting to write in 
"environment passing style".   There's a trivial transform from queries 
that contain XQUF-style updates to a program in "straight XML" style but 
it involves threading through an extra parameter to every function and 
making heavy use of tail calls.   But, those awkwardnesses are, at most, 
just a question of syntax -- no need for "declare updating function".




>
> The pending update list is a way of defining the semantics of the 
> language. We don't expose the formal semantics as XML either, nor the 
> import mechanism, nor many other aspects of our semantics. Most users 
> will use one XQuery engine for a given update, and simply issue an 
> update and observe the results.
>

Updates aren't the *only* kind of extension that need some concept of 
environment passing style.  It's an open-ended design pattern.  Any 
update-specific solution is suspect on programming language theory grounds.

-t