On Sun, Apr 20, 2003 at 04:44:17PM +0100, Christophe Rhodes wrote:
> With the attached patch (which I am submitting for review of the
> sticky bits, and to get it into archives in case of catastrophic disc
> failure :-) I was able to build sbcl on x86/linux using CVS clisp as a
> build host, and on ppc/linux using OpenMCL 0.13.4 as a build host.
> Most of the patch is ensuring that things only get defined once on the
> host, as clisp throws a full WARNING on function or macro redefinition
> (which triggers the FAILURE-P return value from COMPILE-FILE); the
> remainder picks up on one or two actual problems with sbcl. I've
> commented most of the tricky bits (with my initials, CSR, too, for
> easy lexical search). Any comments, particularly on the DSD-NAME
> issue, which I feel should be easily solvable, welcome.
My goodness, look at all the fixes! I think if you encounter too much
resistance from the physics examiners you should print this out, toodle
on over to the CS department and see if *they'll* give you a Ph. D.:-)
> + ;; When we are building the cross-compiler, we could be in a host
> + ;; lisp which implements CL macros (e.g. CL:AND) as special
> + ;; operators (while still providing a macroexpansion for
> + ;; compliance): therefore can't use the host's SPECIAL-OPERATOR-P
> + ;; as a discriminator, but that's OK because the set of forms the
> + ;; cross-compiler compiles is tightly controlled. -- CSR,
> + ;; 2003-04-20
> + #-sb-xc-host
> (when (special-operator-p name)
Would it be possible to use the SB-XC version of SPECIAL-OPERATOR-P
William Harold Newman <william.newman@...>
I was going to correct that to "shows his grammar nazi jackboots off"
but that would be ending a sentence with a preposition.
-- dan_b on <http://tunes.org/~nef/logs/lisp/03.02.05>
PGP key fingerprint 85 CE 1C BA 79 8D 51 8C B9 25 FB EE E0 C3 E5 7C