Does predicate evaluation of for-each equal the result of a choose

I want to validate an assumption that a set of 'or' statements as a predicate of a for-each operates analogous to the choose statement. Either things work as I expect or I just haven't hit the edge case that will cause an issue.

In a choose structure I know that it is evaluated in sequence and when the sequence finds a test that resolves to 'true' the processor drops out of the choose.

   <xsl:when test="number($A) and $A=1">
       <!—do a thing -->
   <xsl:when test="number($A) and $A=2">
       <!—do a thing -->
   <xsl:when test="number($A) and $A &lt; 3">
        <!—do a thing -->

So if the first and third when is true (highly probable), the processor drops out of the choose when the first when resolves as true. This is great and expected. However, when this choose needs to be compressed into a for-each it seems a simple task, but a nagging thought in my mind is that I'm just getting lucky.

<xsl:for-each select="bunch of paths[(number($A) and $A=1) or (number($A) and $A=2) or (number($A) and $A &lt; 3)]

In this example whether the first or third (or both) passes probably doesn't matter, but most of my work operates on financial dates so in that respect I want to be sure that if the first and third both pass that the predicate will evaluate the same as the choose statement.

So, I'm just seeking to ensure that my assumption that has been based on observation of predicate evaluation within the for-each is the same as the choose.

1 answer

  • answered 2018-04-17 04:16 Michael Kay

    A significant difference between the two cases is that with the xsl:for-each example the evaluation context changes (both in the predicate, and in the body of the for-each instruction), while in the xsl:choose case, the context is the same throughout.

    Another difference is that xsl:when always takes the effective boolean value of the test expression (so number 1 means true), whereas predicates have different semantics if the value is numeric.