John Hayes' ANS Forth test suite contains tests that look like this:
: WEIRD: CREATE DOES> 1 + DOES> 2 + ;
WEIRD: W1
W1
I'm rather at a loss as to exactly what this is supposed to do. The ANS Forth Specification on DOES> is largely impenetrable.
From reading the test suite, it looks like it's expecting the first call to DOES> to modify W1, but then that calling W1 activates the second DOES>. I assume the second one operates on the word defined by the most recent call to CREATE, but that's already been DOES>ified, so I'm not sure what that's supposed to do.
gforth passes the test suite, so the tests do seem to be valid; but my pet Forth interpreter doesn't, and I need to figure out how to make it work...
The second call to DOES> also modifies W1.
WEIRD: creates W1 with a runtime action of 1 + DOES> 2 +. The first call to W1 sets the runtime to 2 +.
This is more apparent if you change the code to print something, e.g.
: weird: create does> drop ." ONE" does> drop ." TWO" ; ok
weird: w1 ok
w1 ONE ok
w1 TWO ok
w1 TWO ok
w1 TWO ok
The explanation for this is that DOES> always operates on the latest defined word.