[MLton] mingwows

skaller skaller@users.sourceforge.net
Fri, 11 Nov 2005 15:41:39 +1100

On Thu, 2005-11-10 at 14:58 -0800, Stephen Weeks wrote:
> > I support the idea of an mlton executable - I have never understood
> > why launching a compiler should require two steps instead of one.
> The reason we do it is to have a file that can be easily edited by
> hand and contains all of the platform-specific paths and options.
> This allows people to easily adapt a MLton installation to their
> situation without having to recompile MLton (which is, relative to
> editing a script, quite painful).
> > If you choose to use a driver, it might as well be a compiled SML
> > program rather than a platform specific script.
> This does seem like a reasonable way to go, as compiling a small
> MLton/SML program is not that much more painful than editing a script.
> We'll consider as a way forward after this release.

Yeah it is -- bootstrapping problem. to compile that script
you need a working MLton, but you only want to compile it
if you don't have a working MLton.. :)

Is it that hard to use bash under Cygwin/MSys/Unix,
or if that will not work, use bat files?

Bat is perfectly adequate I should think, although it
is a bit hard for me to tell -- all my real scripts
are written in Python, which runs fine on all platforms.

I have been wrestling with this same problem for
Felix. However .. I'm trying to eliminate BASH not batch!!
Bash is almost impossible to use -- the tokenisation/lexing
substitution rules are beyond me. It just isnt designed
to handle spaces in filenames .. :)

John Skaller <skaller at users dot sf dot net>
Felix, successor to C++: http://felix.sf.net