Re: clisp vs. SBCL

[Chavdar Ivanov <ci4ic4%gmail.com@localhost>]

On Mon, 9 Aug 2021 at 08:45, Yorick Hardy <yorickhardy%gmail.com@localhost> wrote:
>
> Dear Chavdar,
...
>
> Thank you for the information! I also tried with SBCL now and here it is about 13
> times faster than CLISP, and about 4 times faster than ECL.
>
> How did you manage to bootstrap with ECL? I had to patch SBCL as below
> (it is a bit messy, especially the libffi part).

My mistake, I didn't. I had all lisps - abcl and ecl included -
installed, but apparently a long time ago I have used clisp as the
initial lisp to get then SBCL - and kept various SBCL packages to use
as bootstrap compilers for later versions - even on different systems.
I tried your patch, but could not make ecl carry on the bootstrap - I
got a complaint  about the parameters to ecl allowing for an execution
from /dev/stdin (the patch from the mail didn't apply, so I manually
edited the files according to the patch and may have made some
mistake).

I then abandoned that and rebuilt SBCL - now at 2.1.7 - using abcl;
this took hours (understandable, abcl is Java...), but eventually
finished; the resulting executable apparently worked, but the SBCL
tests immediately failed. So I at the end rebuilt SBCL 2.1.7 using the
existing 2.1.6 (adding --fancy to the make.sh parameters in the pkgsrc
Makefile - to enable SB-THREAD among others); the tests now passed
almost until the end - there was an error about exceeding the maximum
file descriptor count at the end. The rebuilt maxima was also
successful in running the tests, even about 5 seconds faster than with
SBCL 2.1.6.

>
> --
> Kind regards,
>
> Yorick Hardy
..

Chavdar