upcoming release

[Matthew Fluet]

> > Having not kept up with all of the runtime, MLton structure, and
> > script changes, I've not been very successful at bootstraping.
> 
> I am curious to know what failed when you attempted to build the new
> MLton from an older version.  That is supposed to be painless.

Mostly it's just been successive changes in Makefiles and bin/mlton script
that haven't been playing nice with other scripts I have to build on
remote machines and with the directory structure I like to work with.

The big issue is that I use one mlton script for all the versions I've got
installed; I use the SML/NJ "trick" of sym-linking to this script and then
extracting the name of the command being run to determine which directory
to set as the root.  (This also mostly solves the problem of being 
able to copy sources and executables to a remote /tmp directory
and not need to sed any of the mlton scripts to adjust for the
move.)  Anyways, various changes to the src/bin/mlton script have been
incompatible with what was my stable version of the compiler.  Variously,
I also tend to do a make clean before doing a self-compile, which has the
side effect of trashing the build directory, so the src/mlton/Makefile
decides to use stubs; but that produces a slower version, even if I've got
a "current" compiler.

Anyways, moral of the story is that I don't have any major hangups with
the way in which builds are being done, but I may need to rethink some of
my other scripts to be more robust in the face of changes.