[MLton] MLton library project

[Michael Norrish]

Hi,

I'd be strongly in favour of what Vesa has suggested, and might be in
a position to contribute some code as well.  (There are some very
simple functions that you might expect to be in the Basis Library, but
aren't.  For a really simple example: the function that flips the order
type so that GREATER goes to LESS and vice versa.  And if C can have
qsort in its standard library, why don't we have a sort in ours?)  

I'd also suggest that the new code be reasonably portable, so that
other implementations are thereby encouraged to adopt the new stuff as
well.  Even if a mlton implementation ends up using a binding to C
code underneath, it would at least be nice to have libraries that
could in theory be implemented in SML.  Contrast for example the
Polyhash structure in Moscow ML, which is built on top of a
polymorphic hash-function of type 'a -> int.

I'm also strongly in favour of the new code being distributed with
mlton if possible.  As Vesa says, this should make it easy for people
to use the library, and may also help focus implementation effort on
what generates the most interest and demand.

But, the really crucial question is what the new library should be
called!  If we already have the Basis Library, perhaps the thing that
builds on top of it could be called the Extension Library.  (That's my
2c suggestion, generated after 2 minutes' thought, anyway.  I'm sure
we can collectively come up with something better.)

I also think Vesa's comments about process and organisation are
clearly the right way to proceed if possible.

Michael.