Re: Several rules vs a single rule with a choose opera

>Trevor Nash wrote:
>> My advice would be to choose the method that is going to be easiest to
>> change - probably the separate templates route.  As to performance, it
>> is true for most software systems that time spent getting the overall
>> design right is much more valuable than time spent on the minute
>> details of implementation.  In short, stop worrying about it.
>
>Thanks for the advice, Trevor
>(as well as for those I have deleted (but not forgot)).
>
>> >  <xsl:template match="node-test[condition2]" priority="2">
>
>> You only need the priority if condition1 and condition2 are not
>> mutually exclusive.
>
>Yes, they need not be mutually exclusive.
>
>> In which case I wonder why you are not testing
>> the common part in the 'choose' method?
>
>Hmm... Because there is no need. :-)
>The two approaches I presented are equivalent in all cases, aren't they?

Logically equivalent yes, but here we are considering performance.
You probably should not need to know - but there are techniques for
coding conditionals based on formula rewriting and statistical
analysis of the input data.  I have only ever used them for compilers
and the guts of an operating system, so its probably off-topic for
this list.  But if your conditions are expensive to evaluate you may
want to think about expressing them in a different way (and a
different order) so that on average the answer comes out quicker.

Regards,
Trevor Nash
--
Traditional training & distance learning,
Consultancy by email

Melvaig Software Engineering Limited
voice:     +44 (0) 1445 771 271 
email:     tcn@xxxxxxxxxxxxx

 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list