[MLton-devel] yeah, that did the trick

Stephen Weeks MLton@mlton.org
Mon, 16 Sep 2002 11:23:22 -0700

> Ok, I've put together a patch file, a binary distribution, and the
> output from the regression script:
>    https://berzerker.nightmare.com/~rushing/mlton/
> The build process is still a bit manual, for example I had to make a
> fake 'self linux' hostmap file.  For now it thinks it's hosted on
> linux.

I installed your tgz on a fresh FreeBSD 4.6.2, and whenever I run it,
I get "mlton: cannot open /etc/ld.so.conf".  This makes sense, because
it thinks it's running on linux.  Can you explain why it works for
you?  I assume you just faked the file.  In any case, I've gone ahead
and added a new MLton.hostType in the basis library MLton structure
and a new Control.hostType in the compiler sources.  Now, when
starting with an SML/NJ compiled MLton on FreeBSD, you should be able
to create a new cross host using the add-cross script, and cross
compile a FreeBSD mlton using the -host flag.  Also, I changed
main.sml so that on FreeBSD it doesn't look for ld.so.conf.

> Most of the patches aren't a big deal:

I checked in all your changes except for the ones to the Makefile and
regression script.  For the Makefile, I've replaced all uses of cp
with $(CP), which expands to "/bin/cp -fpR".  I think this works on
Cygwin, FreeBSD, and Linux.  As to the regression script, I'm not sure
what to do.  You replaced "make -W $f" with "touch $f && make $f".
Maybe it would be better to replace the make's with gmakes?  I also
noticed I needed to do that when making the runtime. 

This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
MLton-devel mailing list