[MLton] Henry's comments on the User Guide

Stephen Weeks MLton@mlton.org
Thu, 12 Feb 2004 14:48:20 -0800

> Re the example on page 3 of -link-opt, I would say something like:
>         If, for example, libgmp.a is in /tmp, then add
>             -link-opt -L/tmp

Excellent.  Done.

> The  stop  runtime switch is fine with me, as long as there is some way to do
> it.

It was easy, and has already been checked in.  I actually quite like
that one can now compile with "-runtime stop" and get an executable
that does no @MLton processing.

> Re the runtime `stop' option, the only  argument  for  -no-load-world  is  to
> allow  someone  to  specify  run-time  options  (like  heap  size) on set-uid
> executables, which is probably worth something.

Yeah, that's enough of an argument for keeping it.

> re _import and _export, I would think it would still be a bit better  to  use
> some brackets other than paren's, since parens are so over-used (for grouping
> and tuples) already, but that is only a very minor feeling.  The switch  from
> semicolon-terminated  to  parenthesized  would  be a big improvement I think.
> The problem with semicolon is that its usual use has a VERY low precedence if
> you  will, so you get used to scanning for them first to break things up, and
> with _import and _export you are pimped.
> Any way, I agree that the incompatibility and the change is bad.  It might be
> worth asking the big users, like the OpenGL and Gtk people.

The low precedence was why I chose semicolon in the first place.  I
didn't want to put anything at all, but was forced to by parsing
considerations.  So, I chose something that was visually unobtrusive.

I'm undecided on the issue.  For clarity, here are the choices:

A. _import "f": string -> unit;
C. _import "f": (string -> unit)
B. (_import "f": string -> unit)
D. _import "f": [string -> unit]
E. _import "f": string -> unit end
F. _import "f": {string -> unit}

I have listed them in decreasing order of my preference.

Does anyone else have any opinions?

> Any way, looks good.  When is the release scheduled for.

I am hoping to do it next week.  There haven't been any serious
problems for a while.  All I have left is a few more tweaks trying to
speed up the elaborator, and possibly some more cleanup of the
-show-basis output.