On Tue, Mar 19, 2002 at 03:45:10PM -0500, John M. Adams wrote:
> I'm trying to determine the best free clos platform. I've played with
> cmucl 18c and found its compiler warning behavior to be distracting.
> Clisp is nice but is missing advanced features.
> How is sbcl in the bogus warning department when compiling clos
> programs? Since the term `bogus warnings' actually appears in the
> cmucl faq, I assume I can leave that undefined here.
I'm sorry, I don't know where to find the current CMU CL FAQ, and
I haven't used CMU CL's CLOS for years, so leaving it undefined
leaves me a little confused.
SBCL should be pretty easy to build if you already have CMU CL
running. You mentioned that you've already played with CLOS in CMU CL,
and if you kept your test cases around, it might be pretty easy to run
them through SBCL and see how much trouble SBCL causes. But I'll also
try to guess and summarize.
A few bogus compiler warnings have been fixed in SBCL, but I don't
know whether they'd be the ones you want.
SBCL has a few other CLOS fixes as well. E.g. in SBCL the code
compiled inside a method automatically knows the type of the argument
if the method was specialized on that type, so it can be compiled
SBCL and CMU CL both have a deep nonANSIism, bogusly distinguishing
between CL and PCL versions of things like CL:CLASS vs. SB-PCL:CLASS
and CL:FIND-CLASS vs. SB-PCL:FIND-CLASS. I've used CLOS a lot in an
application I'm writing, and I've hardly ever run into this problem,
but it's a nuisance for some kinds of programming. The sbcl/TODO list
mentions a possible approach to fixing this problem in SBCL, but
you'll notice that a lot of stuff appears earlier in the TODO list...
> Oh, and how do I change the current working directory in sbcl? Cmucl
> has something like unix:unix-chdir but that doesn't seem to be there
> in sbcl. I'm used to :cd in acl.
Currently SBCL mostly tries to just do the ANSI
*DEFAULT-PATHNAME-DEFAULTS* thing, which mostly overrides the
Unix-level default directory. If there's some other kind of behavior
that you need, and it seems to be consistent with what ANSI specifies
as the behavior of *DEFAULT-PATHNAME-DEFAULTS*, you could report it as
a bug or feature request.
William Harold Newman <william.newman@...>
"Of course, if I dig my house foundations by biting the earth while banging
my own head with a spade, then upgrading to mechanical digger maybe won't
help..." -- Graham Perkins <gperkins@...> in comp.lang.eiffel
PGP key fingerprint 85 CE 1C BA 79 8D 51 8C B9 25 FB EE E0 C3 E5 7C