[MLton] can mlbasis rename the top-level?

[Matthew Fluet]

>> The only thing this doesn't resolve to my mind is whether to use {<{
>> }>}, or _sml _end, or other tokens not alllowed in SML.
>
> My opinion is that _sml _end is that a) the use of _ isn't consistent with 
> our other uses to inicate a non-standard primitive; and b) they are too close 
> to "normal" keywords, that, to me, they seem more out of place than symbolic 
> delimiters.
>
> I'm not wedded to {<{ }>} and would be happy to hear other suggestions.

Another thought: we could push even closer to the 'here-document' approach 
and allow the opening/closing delimiters be chosen on a per-instance 
basis.  For example, I don't imagine it is that much harder to allow an 
opening delimiter that matches the reg-exp:
  {<+{
and a closing delimeter that matches the reg-exp:
  }>+}
with the additional requirement that the number of < in the opening is 
equal to the number of > in the closing.  Now you can always find an 
opening/closing pair that doesn't conflict with any finite amount of SML 
code.