[xquery-talk] XQuery & BEA

[Michael Kay]

> The bad news, 
> however, is that XQuery is not yet a standard - it changes 
> every few months.  To ship a product, one therefore must pick 
> a snapshot of the language, build it, stabilize it, QA it, 
> etc.  Someone commented that our being on an older version of 
> the XQuery spec must mean that our XQuery engine is 
> "pre-alpha" and not trustworthy.  Not so - in WLI, for 
> example, we tracked all XQuery spec changes up to the version 
> released 11-12 months ago - then we decided to stabilize on 
> that version of the language for our alpha, beta, and GA - 
> and to build out all the facilities around it (transformation 
> editor, XQuery source editor), to do QA on it, and so on.  We 
> plan to upgrade to XQuery 1.0 as soon after it's an official 
> spec as possible.

Let me add that we can say exactly the same thing about the support for
XQuery in Software AG's Tamino 4.1. It is possible to release an
industrial-strength product based on specifications that are not yet
finalized, but it is not possible to track every revision to those
specifications, and still deliver an acceptable level of stability to
users. Tamino 4.1 came out in January 2003, based on the August 2002
specifications, and although we may add some features from later working
drafts in subsequent versions, it is not our intent to track every
single change in the working drafts: the primary goal is to deliver a
conformant and high-quality XQuery 1.0 implementation as soon as
possible after a Recommendation is signed off.

This is the opposite of my strategy for Saxon, where I do try to track
the changes to specifications as closely as I can, and I specifically
warn users that they are on the bleeding edge and should expect a rough
ride.

Michael Kay
Software AG