On Sun, Dec 17, 2000 at 02:42:01PM -0500, Nathan Froyd wrote:
> I downloaded sbcl-0.6.9 the other day and attempted to compile it with
> the binary release of CMUCL 18b. This is on a RedHat 6.1 machine. The
> last part of the output from the compilation process is:
> Error in function C::NORMAL-READ-ERROR: Read error at 81742:
> "#+cmu (stub:probably/\ similar but has not been tested))"
> Looking at the source file, it appears that this is some part of ugly
> hack to get things to compile correctly. Unfortunately, it also looks
> like to get sbcl compiled in the first place, one has to have a working
> version of sbcl. Any thoughts on fixing this?
I'm sorry about the screwup. Obviously I realized when I wrote that
code that there would be a problem with bootstrapping under CMU CL,
but I forgot to go back and address the problem. Further, I didn't
bother to test using CMU CL as cross-compilation host when I released
sbcl-0.6.9: shame on me!
to fix this:
Check out version 0.6.9.7 (checked in on Dec 14) from SourceForge
CVS. (You can follow the "CVS Repository" link from
<http://sourceforge.net/projects/sbcl> if you don't already
know how to do this.)
on needing SBCL to build SBCL:
I'd like to be as far from that situation as practical. I'd be
particularly happy if CLISP became ANSI-compliant enough to serve
as a cross-compilation host. Failing that, though, CMU CL is the
only other open source Common Lisp which is ANSI-compliant enough.
And as you can see from the problem and the diffs to fix it, even
supporting CMU CL requires some workarounds.
Yes, it's an ugly hack. It's to work around some problems with
non-ANSI EVAL-WHEN which are common to both SBCL and CMU CL.
A truly ANSI EVAL-WHEN in SBCL isn't going to happen until I
can get rid of the IR1 interpreter, so that I can start punting
the special cases throughout the system which are required to
support the IR1 interpreter. That should happen sometime in 0.7.x;
and 0.7.x should happen after fixing more of the more glaring
problems in the 0.6.x BUGS list.
William Harold Newman <william.newman@...>
PGP key fingerprint 85 CE 1C BA 79 8D 51 8C B9 25 FB EE E0 C3 E5 7C