David Allsopp wrote:
> Brian Hurt wrote:
> 
>> Jon Harrop wrote:
>>
> 
> Agreed - then we could also have more sensibly located functions such as
> Int.of_string (note that it's the same length as int_of_string!!) and remove
> lots of random functions from Pervasives!

Of course we'll just add an extra Int module and leave Pervasives alone, or at
least make it trigger a warning.

> All that said, and especially as StdLib changes are reasonably rare, I find
> having files IntSet.ml and IntSet.mli containing:
> 
> include Set.Make(struct type t = int let compare = Pervasives.compare end)
> 
> and
> 
> include Set.S with type elt = int
> 
> isn't too bad (except that you have to include IntSet.cmo/.cmx when
> compiling, obviously)
> 
> 

I agree, and someone might write an optimized version for ints or floats.

But Xset & Xmap also provide opportunities to add some missing functions.
In particular, I often use Sets or Maps as a good priority queue/heap substitute, and
reverse_fold / reverse_iter come in handy.

The only problem is that I have an allergy to CamlCaseIdentifiers but I'll
just swallow some antiHistaminics.

-- 
Berke DURAK