That's it: an infinitely looping foreach isn't a foreach. It's not a foreach in perl. It's not a for-each in standart scheme either, as r5rs clearly says that the arguments to for-each are lists. (In some interpreters, you can give an infinite list to for-each, but that's clearly an extension. Others, like mzscheme, raise an error on an infinite list.) For in sh and foreach in csh won't allow you to loop infinitely (except for the new arithmetic for loop in bash, but that's not a foreach loop, it's rather a c-style for loop).

But you said Turing complete.

A Turing machine is a theoretical construct that needs an infintely long tape, so any implementation has to be a subset, without invoking religion (-: and then you first have to prove that the g.d you use really exists :-).

So isn't it a bit non-gracious to complain about an infinite list?

I mean, it wouldn't be impossible to use a tied array. (The semantics for how a list is created makes that impossible, I think?)

(define a (list 1 2 3))
(set-cdr! a a)
(for-each (lambda (x) (display "hello world\n")) a)
[download]

Please read what I said in the parent node.

That's not a proper list, so r5rs doesn't guarantee this will work. It does work with some scheme implementations, but MzScheme says

for-each: expects type <proper list> as 2nd argument, given: #0=(1 . #
+0#); other arguments were: #<procedure:STDIN::59>
[download]

if I type this code in it.

Update: let me quote from the r5rs.

Read more... (1362 Bytes)