[Axiom-developer] New to this list

[Jaap Weel]

I have a bunch of things to say, but haven't typed everything up yet, 
so let me start with what I have:

First of all, congratulations on getting the Axiom project up and 
running. Clearly the design philosophy that Tim is proposing for Axiom 
is do the Right Thing even if it takes time. A very rare perspective 
indeed, these days. I appreciate that much. At the moment I'm compiling 
Axiom on an x86 Linux box downstairs from me. Seeing how things are 
going, it seems I should soon be able to run it on my own Mac and/or 
DEC.

- Mac OS X, case issues: this goes wrong right when I get the CVS. When 
I check out the files, it just refuses to check out both 
FOO.input.pamphlet and foo.input.pamphlet. It seems that the 
.input.pamphlet files are the only files that cause this problem. Is it 
possible to just rename these file, like foo.* and big-foo.* or so? Or 
stick them in a different directory?

- Mac OS X, build issues: as the only build instructions in README 
involved /spad/mnt/linux, I just literally copied them to see what 
would happen. I know GCL hasn't been ported yet, but I thought I'd see 
how far I'd get.

[japie:/spad] weel% sudo bash
Password:
bash-2.05b# export AXIOM=/spad/mnt/linux
bash-2.05b# export PATH=$AXIOM/bin:$PATH
bash-2.05b# make

/spad/int/lib/fnct_key.c: In function `handle_function_key':
/spad/int/lib/fnct_key.c:308: `SIGCLD' undeclared (first use in this 
function)
/spad/int/lib/fnct_key.c:308: (Each undeclared identifier is reported 
only once
/spad/int/lib/fnct_key.c:308: for each function it appears in.)
make[3]: *** [/spad/obj/linux/lib/fnct_key.o] Error 1

In BSD-based systems such as OS X, SIGCLD is called SIGCHLD. I would 
gladly start patching the code to deal with this kind of stuff, but I'd 
like to know if there is any policy about conditionalizing code. Can I 
just use a couple of #ifdef c.q. #+'s?

- Common Lisp implementation. There was concern from the French side 
about the speed of the underlying Lisp. The fastest free Lisps are 
probably CMUCL/SBCL. I am specifically mentioning SBCL here. It is very 
similar to CMUCL, but it is easier to port. This is because CMUCL needs 
CMUCL to build, whereas SBCL will build with another base lisp -- 
thanks to Christophe Rhode-- such as dreadfully slow but hyper-portable 
CLISP. There is, among others, a port to OS X. I just installed the 
binary and it's blazing (almost 2x OpenMCL on a simple test case).

For benchmarks, see http://lispm.dyndns.org:8000/ , 
http://www.cliki.net/Performance%20Benchmarks , 
http://www.cliki.net/Performance%20Benchmarks , 
http://www.bagley.org/~doug/shootout/craps.shtml .

I like the idea of compiling against more than one lisp. Various 
implementations each have their advantages, that the advanced Axiom 
user should be able to choose from. Also, it will eliminate non-ANSI 
code, as has been seen already with ONEP. More about ONEP later. At any 
rate, I think SBCL may be a little cleaner and more portable to use 
than CMUCL.

========================================================================
Jaap Weel                   Campus address:        | dorm (626) 795-9748
Caltech, Blacker '05        Caltech MSC #874, Pasadena, CA 91126, U.S.A.
www.its.caltech.edu/~weel   Permanent address:     | home +31-46-4337033
E-mail: address@hidden    Kelderstraat 2-4, 6171 GB Stein, Netherlands
========================================================================