Yeah, I found the family of functions that ‘every’ belongs to as a result of Zak’s response. Hurrah for the web!

But yes, if I understand you correctly, (null x) and (not (eq t x)) are not quite the same. Since the second form does what I said I wanted whereas the first returns true for any non-nil value.

Thanks for the LispCast stuff, by-the-way!

Steve

At first you say this:
>return true if all the elements are T and nil if any are not T

Then you start counting (null x), not (not (eq t x))

Try this:

(every #’(lambda (x) (eq t x)) list)

(time (eval (cons ‘and (loop for i below 1e6 collect t))))

But it blew up. And comparing it for 1e5 did not seem to show it being faster. Bummer.

No, it isn’t:

CL-USER> (eq ‘t
NIL

You’re right that identity is what I was looking for though.

(loop for item in list always item)

(lambda (x) x) exists already. It is called IDENTITY.

(every #’identity list)

(notany #’null ‘(t t t t t t))
(every (lambda (x) x) ‘(t t t t t t))

Both were slightly slower than the loop, but twice as fast as reduce. count-if and find-if were both a little slower than reduce.

Personally, I think notany is the clearest way to express this. Note also that not all of these forms are semantically equivalent: some test that all elements are the symbol ‘t, and some test that none of the elements are nil. It is traditional Lisp style (and the behavior of CL’s if) that any value that isn’t nil is considered true, so if you’re testing to see if a list of values are true it’s probably best to use a form that does the latter.