[MLton-user] experimental x86-darwin packages
joe.hurd at cl.cam.ac.uk
Mon Oct 30 14:24:25 PST 2006
> I suspect what happened is that John's machine doesn't have gmp installed
> via fink in /sw/lib, but rather has gmp installed in /usr/local/lib.
> There may be some linker magic that one can use to 'relink' against gmp in
> /sw/lib. Alternatively, you could add a symlink in /usr/local/lib to the
> /sw/lib gmp dynlib.
Good idea: I created a temporary symlink to test it out, and
everything worked fine.
I ran an example program of mine that takes about 30 mins, and got
better performance on my Mac with a 1.83 Ghz Intel Core Duo processor
limited to 800Mb than a linux PC with an Intel(R) Xeon(TM) CPU 3.20GHz
limited to 1600Mb. [The memory limitation is derived from the compiler
option -runtime ram-slop 0.4 I use on both systems, and the Mac has
only 2Gb compared to the PC's 4Gb.]
The version of MLton on the PC is built from sources in June 2006, and
both versions use the compiler option -drop-pass deepFlatten because
of the compiler bug that I exposed a while back. By the way, has that
bug been fixed yet? More generally, is there a MLton bug tracking
system so that I can see when my favourite bugs are fixed?
More information about the MLton-user