Yep, section 9.3.3 says: > [A choice] is resolved sequentially, or by direct dispatch. > Sequential choice resolution occurs by parsing each choice branch in > schema definition order until one is known-to-exist.
You're correct about your schema. In the latter schema, the integer branch will never be taken unless something causes the string branch to create a parse error. On 12/24/20 8:28 AM, Roger L Costello wrote: > Hi Folks, > > The C programming language specification says that the order of evaluating > the arguments of a function call is unspecified. So, one cannot rely on the > first argument being evaluated before the second argument. > > Does the DFDL specification explicitly state an order of evaluating the > element declarations in an xs:choice? If so, where in the specification does > it state this? > > Why do I ask? Because this works: > > <xs:choice> > <xs:element name="foo" type="integer" ... /> > <xs:element name="bar" type="string" .../> > </xs:choice> > > but this doesn't work: > > <xs:choice> > <xs:element name="bar" type="string" .../> > <xs:element name="foo" type="integer" ... /> > </xs:choice> > > Why doesn't the latter work? Because a string matches anything, so the foo > element will never be evaluated. > > /Roger >
