afterthoughts Re: [Sbcl-devel] --noprogrammer problem

[William H. N. <wil...@ai...>]

On Wed, May 09, 2001 at 08:22:18AM -0500, William Harold Newman wrote:
> On Tue, May 08, 2001 at 11:07:50PM +0100, Christophe Rhodes wrote:
> > On Tue, May 08, 2001 at 08:04:53AM -0500, William Harold Newman wrote:
> > > On Tue, May 08, 2001 at 10:08:21AM +0100, Christophe Rhodes wrote:
> > > > I have been given a top-level prompt when recompiling the debian cmucl
> > > > graystream package with --noprogrammer; yes, I know that this is an
> > > > odd thing to be doing.
> > > [etc.]
> > > > Only when I explicitly (quit) does the compilation run continue.
> > > 
> > > Yes, that's what --noprogrammer was supposed to avoid. Strange, but it
> > > wouldn't be the first time I made a mistake.
> > 
> > It seems odd to me, on a cursory reading of the code. It seems to
> > occur on compilation of code from defsystem containing references to
> > packages that do not exist. It's not just cmucl-graystream; this also
> > occured with uncommonsql, which used references to lisp::foo when it
> > meant common-lisp::foo...

[snip]

> I.e., for code outside COMPILE-FILE, there's no signal to catch. One
> way to see this is that in the absence of --noprogrammer, the failing
> COMPILE-FILE wouldn't cause you to enter the debugger.

I realized a few minutes after I sent this that that might not be the
case, because of bug 24 (in the BUGS file). And in fact, when I try
  $ sbcl --noinform
  * (compile-file "foo.lisp")
I do end up in the debugger.

Fixing bug 24 is one of the ten or so probably-not-too-hard things I
listed a while ago as wanting to do before 0.7.0. But I probably won't
be dealing with any items on that list for at least a week. Meanwhile,
if you'd like to send a more exact description of how to replicate the
problem you're having, I can at least attach it to the description of
bug 24, to try to make sure that when the basic bug 24 gets fixed,
your problem gets fixed too.

-- 
William Harold Newman <wil...@ai...>
"Tweak alpha so it sends SIGBUS for unaligned access, and does NOT
do a fixup. This encourages people to fix their code." -- a commit
note from <http://www.OpenBSD.org/plus29.html>
PGP key fingerprint 85 CE 1C BA 79 8D 51 8C  B9 25 FB EE E0 C3 E5 7C