On Sun, Jun 20, 2004 at 03:18:05AM -0500, Brian Downing wrote:
> Once thing I noticed when working on some SBCL internals tonight is that
> Slime is compiling things for SBCL with very little debug. This can be
> seen by compiling:
> (defun bar (x y) (zot x y))
> from the Slime REPL and the *inferior-lisp* REPL. The Slime-compiled
> one has a very short disassembly, and tail-calls the FDEFINITION. The
> REPL-compiled one is quite long, and doesn't. This is strange, though,
> since sb-c::*policy* evaluates to the same in either environment.
> While this was sort of nice for me (Slime's low debug made my find and
> fix a nasty stack-destroying bug), it seems bad that Slime-compiled code
> is defaulting to low debug levels, and seems to not be respecting the
> Is this happening on anyone else's installation, and does anyone have
> an idea why this is happening?
Actually, it looks like this might be a problem with the SBCL repl.
Even if I (proclaim '(optimize (speed 3) (debug 0))) at the SBCL repl,
(defun bar (x y) (zot x y)) still has tons of debug in it. Same with