I thought I'd let you know about an interesting processing error that occurred today when testing IBM DFDL.

An expression of the form:

{if (fn:exists(x) and fn:contains('abc', x/y)) then 1 else 0}

gave a parser error saying that x/y didn't exist, when x was not in the message. This surprised me, as I was assuming that the IBM DFDL parser would lazily evaluate the operands of the logical expression. I was going to raise a bug, but thought I had best read the XPath 2.0 spec first. The relevant section is here:

http://www.w3.org/TR/xpath20/#id-logical-expressions

The section says that unless you are in XPath 1.0 compatibility mode (which a DFDL processor would not be), the order of evaluation of operands of a logical expression is actually implementation dependent. So it is therefore unsafe to rely on lazy evaluation to implement a guard in the manner that I had. I have rewritten the expression as:

{if (fn:exists(x)) then if (fn:contains('abc', x/y)) then 1 else 0 else 0}

I think this might catch people out and is certainly worthy of mention in (at least) a tutorial on DFDL expressions.

Regards
 
Steve Hanson
Architect, IBM DFDL
Co-Chair, OGF DFDL Working Group
IBM SWG, Hursley, UK
smh@uk.ibm.com
tel:+44-1962-815848
Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number 741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU