Wesley W. Terpstra
Thu, 16 Dec 2004 08:48:23 +0100
MLton MLTONVERSION (built Thu Dec 16 07:21:03 2004 on paer)
terpstra@paer:~$ uname -a
Linux paer 2.4.26-64-smp #1 SMP Thu Apr 29 04:59:03 UTC 2004 parisc64 GNU/Linux
There are some regression with signals and threads, but mlton can at
least rebuild itself now. The 'real' test also has regressions b/c the
result differs after about the 10th digit; is this ok?
I've switched everything over to C99 (runtime, bytecode, c-codegen,
and constants). I had to tweak the assembler codegen a bit too ('#' is
not a valid comment with gcc -std=c99). The only parisc specific code in
MLton is the datatype = HPPA, everything else is just C99 portable.
Oh yeah, I turned off the profiling stuff; I have no idea what the EIP
register is even called on hppa. The 'gregs' array doesn't exist here.
Anyways, things are working out pretty fine.
I'll mail the patch once I've gotten the regressions fixed and ported
it to a few more platforms. Then we can dupload a bootstrap. =)
Wesley W. Terpstra <firstname.lastname@example.org>