> I thought that the simpler expression below with a decent optimizer
> will be as efficient as the above:
>
>     exists($context[normalize-space()])
>
> as it is sufficient to find the first non-empty text() descendent to
> conclude that the argument to the exists() function is the $context
> element.
>

Saxon does part of this optimization - normalize-space() in a boolean context
is evaluated by scanning the string until a non-whitespace character is found
(effectively rewriting as a function call contains-non-whistespace()).

It doesn't do the other half, recognizing that
contains-non-whitespace($element) can be evaluated as
exists($element//text()[contains-non-whitespace(.)]).

There's always scope for more such optimizations, but there's a law of
diminishing returns. It's not unusual to find that some such optimization
isn't triggered by a single one of the 30,000 QT3 test cases, and then you
have to ask whether it's worth implementing.

Michael Kay
Saxonica