Re: The output of evaluating an XSLT transform is the

De : Bill Michell wrote"

> Of course, it *is* possible to write extension
functions that
> do have side-effects. But this should really only be done
>
mindfully, and in full knowledge of the consequences. And then
> only if you
don't mind the output of your transformation being
> non-deterministic...
Well, having side effects does not mean your transform is
non-deterministic.
But the time relation between two side
effects is non-deterministic.  For
instance, xsl:result-document
can have side-effects, but has been designed to
minimize the
potential problems (e.g. you cannot read a URI that has been used
in xsl:result-document).

  And of course, you have to be sure to understand
the extension
mechanism you use (the one provided by the processor your are
targeting).  Rspecially the relationships with the optimizer ("if
my extension
function does not return anything, can it be thrown
away by the optimizer?",
"if my function is called twice with the
same parameters, can it be
memoized?", etc.)

  Regards,

-- 
Florent Georges
http://www.fgeorges.org/