ARGH!! I committed a stupid error in my last mail.
I said:
> Quoting the functional argument prevents this level of evaluation from
> happening:
>
> sublist( '[1,2,3], 'myfun )
>
> will apply the literal function myfun to each of 1,2,3.
This is true, but since the symbol myfun was bound to evenp, Maxima
actually ends up calling evenp. This is because Maxima looks at the
"value" property of a symbol if there is no "function" property --
even in "apply" context.
This is probably a design flaw (because there is no reasonable way to
apply myfun literally in a clean way), but again that is a different
discussion.
-s