[
Lists Home |
Date Index |
Thread Index
]
I agree with you 100%. Here's a quote [without
permission] from one of our XQuery developers which I
doubt ever bubbled back the XML Query WG.
"Before proposing a solution, I'd suggest establishing
what problems the [XQuery] type system is trying to
solve:
Validating external inputs and/or generated outputs?
Converting between lexical representations and values
(formatting/parsing)?
Eliminating common user errors at compile-time?
Working with structures as first-class citizens (that
can be passed around to functions, etc.)?
Optimizing execution performance?
Proving correctness of query normalizations?
Pruning the set of applicable query normalizations at
any given expression (optimizing optimizer
performance)?
Other things I've not considered here?
I think the current type system confuses these goals
into a mushy complex system of rules, and would
benefit from better factoring.
I also think the current type system actually fights
against some of these goals, instead of supporting
them. To give two examples, I believe the current
dynamic type checking rules slow performance instead
of improving it; and the complex static type
inferencing rules obfuscate user errors instead of
catching them."
Personally I think the W3C XML Query working group is
divorced from reality and dogmatically following the
action items on its charter without actually thinking
through the practicality of the specification they are
building.
However, come when XQuery reached recommendation
status we will ship a kickass XQuery implementation.
Believe that. :)
C'est la vie.
--- Joe English <jenglish@flightlab.com> wrote:
>
> > Queries on large sets of data *do* need to be
> efficient - and when you want
> > to be efficient, you try to avoid actually
> retrieving the data whenever
> > possible. Reasoning about types statically is one
> of the ways we do this.
>
>
> I still don't buy this argument.
>
> Static typing can help optimize low-level operations
> that are
> close to the machine (for instance this is why as a
> rule SML is
> more efficient than Scheme), but languages that do
> heroic
> optimizations -- APL, SQL, and Common Lisp for
> instance -- tend
> to be dynamically typed. Static typing is neither
> necessary
> nor sufficient.
>
> What optimization strategies does the XQuery type
> system enable?
> Has anyone implemented, or even just designed, a
> database access
> optimizer that takes advantage of it?
>
>
> --Joe English
>
> jenglish@flightlab.com
>
>
-----------------------------------------------------------------
> The xml-dev list is sponsored by XML.org
> <http://www.xml.org>, an
> initiative of OASIS <http://www.oasis-open.org>
>
> The list archives are at
> http://lists.xml.org/archives/xml-dev/
>
> To subscribe or unsubscribe from this list use the
> subscription
> manager: <http://lists.xml.org/ob/adm.pl>
>
=====
THINGS TO DO IF I BECOME AN EVIL OVERLORD #79
If my doomsday device happens to come with a reverse switch, as soon as it has been employed it will be melted down and made into limited-edition commemorative coins.
__________________________________________________
Do You Yahoo!?
Sign up for SBC Yahoo! Dial - First Month Free
http://sbc.yahoo.com
|
