Subject: Re: Buggy BinarySearch implementation, real life and a curious soul...
From: rpw3@rpw3.org (Rob Warnock)
Date: Sat, 10 Jun 2006 05:42:25 -0500
Newsgroups: comp.lang.lisp
Message-ID: <F7GdncqweqyMPxfZnZ2dnUVZ_rudnZ2d@speakeasy.net>
Juho Snellman  <jsnell@iki.fi> wrote:
+---------------
| Rob Warnock <rpw3@rpw3.org> wrote:
| [... Python's tagging scheme ... ]
| > Preserving this useful characteristic on a 64-bit machine means
| > using a "three zero bits" lowtag for fixnums there, which is what
| > SBCL appears to have done.
| 
| Yes, that's one reason 64-bit SBCL has a different lowtag size. On the
| other hand, the unfinished CMUCL x86-64 port uses two zeroes lowtag
| for fixnums, just like the x86 CMUCL. I don't know why that decision
| was made.
+---------------

I'm not sure it *was* explicitly made. I suspect that (non)decision
might need to be revisited [to do it the SBCL way] before the CMUCL
x86-64 port gets finished, since the "fixnums index lispobjs" is so
heavily wired into CMUCL...


-Rob

-----
Rob Warnock			<rpw3@rpw3.org>
627 26th Avenue			<URL:http://rpw3.org/>
San Mateo, CA 94403		(650)572-2607