[MLton] r4723: Initial commit of the extended basis library

[Vesa Karvonen]

Quoting Stephen Weeks <sweeks at sweeks.com>:
> I wonder if using a suffix of "Iso" on the isomorphisms,
> e.g. "Char.intIso", would be a nice convention.

Perhaps.  Having a suffix has the side-effect that one can then specify
an "isomorphism" between arrays and vectors without having to respecify
the vector value that that basis array signatures already have.  So,
I've made the change.

> Which is preferable, "isZero" or "is0"?  For some reason, the latter
> grates on me, possibly because I am used to seeing integer suffixes
> distinguish related variables (x0, x1, x2, ...).

I thought about this myself and ended up with is0 earlier.  Perhaps
isZero is more customary (lesser surprise), so I've made the change.

> In MkIntegerExt, the definition of isEven, it would be faster to use
> "rem (i, fromInt 2)" than "i mod fromInt 2".

Changed.

> I saw the "svn:eol-style native", which we haven't been using.  It
> seems like a good thing, though.  Does anyone know of any problems
> with it?  Should we make it convention to put it on all files?

AFAIK, it is a good idea to have it on all text files that should use
the native eol style.

Thanks for the comments.

-Vesa Karvonen