Christophe Rhodes <[EMAIL PROTECTED]> writes: > Certainly, setting SB-C::*CHECK-CONSISTENCY* to T, then evaluating the > tagbody form in question, in a recent CVS sbcl gives no errors and a > NIL return value, as I would expect.
That's true, but even CVS SBCL, updated the length of an SBCL build ago, will show the inconsistency when COMPILE-FILE the code below. (defun f () (tagbody (go loop) (BLOCK G1472 #'(LAMBDA () (RETURN-FROM G1472 NIL))) loop)) Maybe DELETE-BLOCK should somehow arrange for deleting the entries created in the IR1 translators for BLOCK and TAGBODY? Don't know. It would probably be nice.
