[xquery-talk] [ANN] presenting XQuery - GSD language at MarkLogic User Group London Nov 24

James Fuller james.fuller.2007 at gmail.com
Wed Nov 9 02:25:31 PST 2011

On Wed, Nov 9, 2011 at 11:13 AM, Andrew Welch <andrew.j.welch at gmail.com> wrote:
> Hi all,
>> But are they gonna learn a new programming language because it is
>> allegedly more productive? No way...
> Is there a killer app yet?

unsure what you mean, is there a killer app for SQL ?

the nonproblem of xquery is that it has a much wider applicability then SQL.

>> It seems to me that for XQuery to go mainstream, we need to go in
>> areas where XQuery is not just more productive but simply
>> indispensable.
> Playing devils advocate:
> - XQuery is only productive because of proprietary extensions for a
> given product

sure but this is the case with every nascent technology,

if XQuery hits adoption then it will require some weeding to bring
core XQuery in line ... there are the EX- type standards ... which btw
is thankless work that everyone wants but no one wants to spend the
energy to actually contribute to.

I whole heartedly believe in the standards but I also whole heartedly
believe in going 'exotic' if there is a pragmatic need that has not
yet been addressed.

> - XQuery on its own isn't a single-tier, do-everything language

XQuery is first and foremost a dsl for data .... I don't buy into the
XRX thing because it presupposes that we live in a homogenous world.
Even if this is the case v2 of whatever data format will ensure

I think XQuery main strength is as a query/update language which has
modular extensibility in terms of how its specified ... its a
replacement for SQL and thats where we should be aiming.

The ability to write full blown apps in it is wonderful and something
I do everyday with it, but I feel that we first need to gain adoption
as a query language, which means we need to figure out the story of
working with relational databases in a standard way.

First port of call is probably tactical e.g. standardise interaction
with RDBMS via some ODBC like interface.

> - Each vendor has effectively created a dsl.

sure, but they are all related and its not like vendor lockin of 20
yrs ago e.g. the problems of porting are far from intractable and I
would argue that we need a whole lot more adoption to warrant
standardisation ... core xquery across most of the vendors is pretty
consistent though you maybe surprised by things like context and
implicit typing ... this is more to do with the imprecision of the
XQuery specs then anything.

most of the lockin revolves around module libraries versus the xquery
language getting doped up with something else.

Jim Fuller

More information about the talk mailing list