MLton vs. OCaml

[Henry Cejtin]

re  -DMLton_safe=0, is that really fair?  I.e., isn't OCaml really equivalent
to default safety in MLton?

I'm very confused.  You say that x_2379 is `main  loop  of  ray  tracer'  and
`line  825  of  render.sml'.  There isn't a render.sml, and render.fun is 231
lines long.

The fact that the flattener flattened too much (21 args) is a  real  problem.
Clearly  there  are lots of times when exactly this flattening would be good.
I would think that the only time it was really bad was if:
    All calls unpack these arguments from other tuples (so re-partitioning it
        and passing the tuples as a tuple would have been a big win.
    All the arguments are passed around a lot.
Was this the case?

OCaml has syntax for loops?  I can see the attraction for array subscripting,
although I would just as soon not have it, but  for  loops?   You  mean  like
`while', which SML stupidly has?

I  have  to  admit that on the occasions when I have tried to read some OCaml
code, mainly off the net, the syntax doesn't look good: no  better  than  SML
(which  is  pretty  bad)  and  yet  different.   (Things  like fun vs. fn for
lambda.)