Re: Whither MrFlow?

[Matthias Felleisen <matthias@ccs.neu.edu>]

Lauri, 

1. I have 10 years of experience with non-graphical type inference
   both HM and SBA. It is clearly the worst thing that anyone in 
   CS ever came up with. 

2. I was also an un-fan of graphical gimmicks. 

3. The key is to give good feedback when MrX sees type errors. 

   The best explanation anyone ever came up with is by showing 
   a counter-example value and its path to the problem site. 
   If you can tell us how to show this path w/o going graphical 
   but in a sensible manner, implement it. 

4. The "assertion" f : T -> S is not at all what you expect from 
   ML. It means 'add T to all possible inputs to f. Then check that 
   S is the only kind of output that f produces.' 

   You really want to say something like (forall (x) (x -> x)) (I think
   this is the proper syntax.) But even then you're saying f will get all
   kinds of values PLUS strings. 

-- Matthias