[Axiom-developer] sbcl port

[Waldek Hebisch]

At:

http://www.math.uni.wroc.pl/~hebisch/prog/sbcl4c.diff.gz

you may find my current patch allowing compilation with sbcl.
The patch shoul apply to current (revision 534) wh-sandbox.

Currently build requires some manual intervention:

1) I ommited diff to interp-proclaims.lisp, one has to replace it
   by safety declaration:

(eval-when (:execute :compile-toplevel :load-toplevel)
  (proclaim
    '(optimize (safety 3))))

Alternativly, one can use empty file.

2) Build dies during tests due to lack writablep function.

3) The reclos test do not finish, it looks that 

relativeApprox(squareDiff8,10**(-3))::Float

line goes into infinite loop.

Also, currently sbcl spends a lot of time (more than 1 hour)
compiling the function jordanAdmissible? from
AlgebraGivenByStructuralConstants.

Things that do not work:

1) I fake sockets, so that graphics tests can go on, but you
   get no output.

2) Output that should go to standard output goes to the terminal
   so redirection does not work (I use the script program to
   capture build logs).

3) Due to 2 I do not know if tests work: with the exceptions mentioned
   tests go forward but I did not check results.
 
There is some quick and dirty code:

- I hacked src/lisp/Makefile.pamphlet so it works for me, but as
  I wrote in other message we should load support code here.
- sockets are fake
- not-a-number constant is fake (Axiom usage looks bogus and we
  probably should remove not-a-number constant, but ATM I just
  fake it)
- almost empty interp-proclaims.lisp (we should have real
  proclaims)
- I had to define posix-chdir in the first image because later
  images were unable to load posix package

and probably more.

Following Juregen Weiss in few places I test functions for equality
like:

f = FUNCTION newGoGet => SUBST('_$,domname,devaluate CAR r)

This code may fail because Lisp may produce fresh function each
time we execute the test.

The is also some clisp code: it is probably not working yet (and
it uses paths corresponding to locations on my hard drive...).

-- 
                              Waldek Hebisch
address@hidden