[MLton] building mlton on Solaris 10

[Matthew Fluet]

>> So, I'm still confused by the fact that you are producing the expected
>> sizes for sparc-solaris, bt the regression script is comparing against
>> mlton.share.ok.
>
> Vaguely I remember that in the beginning the regression script did not
> find bin/platform for some reasons, that could explain this.

That would explain it.

> I'll do the regression again once I get a working 32-bit gmp on my
> Solaris box. Unfortunately the one on sunfreeware.com is 64-bit. I'm
> getting error messages about bad so's.

Your previous mail suggested that the only failing regressions was 
mlton.share.  But, many of the regressions use the IntInf structure; 
hence, would need gmp.  If they compiled and passed, then there must be a 
32-bit gmp on the box.