XQJ (JSR 225)

[Per Bothner]

On 06/25/2009 09:46 AM, Ken North wrote:
> Michael Kay wrote:
>>> The difference between them is that XQPreparedExpression contains a compiled
> expression or query, whereas XQExpression merely holds the dynamic context
> for executing an expression or query
> .. Whenever I complained about the design (or even the nomenclature) I was told
> it would all seem completely intuitive to people who know JDBC.
>
> That's true. It has its origin in the difference between dynamic and static SQL, with a prepared query being an attempt to achieve some of the performance benefits of static SQL (compile and optimize repetitive queries).

The problem is the design is so terribly non-orthogonal.

It's ironic how relational database theory is all about
"normalization", which is is like orthogonality for data.
The way JDBC/XQJ handles "statements"/"expressions" fares
pretty badly in that respect.

Unfortunately, there was very little interest in clean or
elegant API design.  Conceptual compatibility with JDBC,
however broken, was very important; compatibility was JAXP
was less important; compatibility with java.util.Collections
or javax.script (JSR-223, which started about the same time,
but finished in time for Java 6) even less so.

But I think the biggest problem was a lack of resources,
especially after IBM pulled out.  Improvements and cleanups
that weren't critical weren't doable - the attitude was:
"if people can get their work done with the API as-is then
it will have to do" - which is understandable, if unfortunate.

I couldn't meaningfully contribute in that environment, and
I didn't want my name on what it was turning out to be,
so I resigned from the expert group.  (In honesty, another factor
was I had gotten very little interest in my XQuery implementation,
so it was hard to justify spending much time on XQJ.  In
contrast, there are many people using my Scheme implementation.
(That is Scheme, as in the Lisp family, not Schema.))
-- 
	--Per Bothner
http://x-query.com/mailman/listinfo/talk   http://per.bothner.com/