[MLton] mlbs and soft and hard links

[Stephen Weeks]

> Re  speeding up MLton type checking of large programs, do you think that that
> is really do-able without changing the order of source code?  I.e., given mlb
> files, I can see (I think) how I might structure a program so that a bunch of
> files are compiled in the same environment (i.e., no one of the files results
> is used in the compilation of any of the others).  Given that, there is a lot
> of freedom to re-order compilation so that the most  recently  changed  files
> are  compiled  last,  or  so  that the context used in compiling the recently
> changed files is long-lived.  That way you will spot errors very quickly.
> 
> Is this what you have in mind?

No.  The proposal I sent does not do any re-ordering.  It simply
caches, at hopefully intelligent places.