Re: Overloaded functions in Ocaml?

On Wed, 2 Mar 2016 09:41:47 -0600 "Douglas Lewit [email protected]
[ocaml_beginners]" <[email protected]> wrote:

> Hi everyone,
> 
> Quick question about functions in Ocaml.  Are there any "overloaded"
> functions in Ocaml?


No, there aren't.

> So I guess my question is this.  Is there a way to create a more
> generic Ocaml function that could generate a list of integers OR
> generate a list of floats depending on the arguments that I provide?
> There probably is I'm guessing, but I have no clue how to approach
> that.

Yes, you could either write a stub in C to do the job or use the Obj 
module (you'll need Obj.is_block Obj.tag and Obj.double_tag). But I 
wouldn't recomend implementing such a function. Its type would get 
inferred to 'a -> 'b -> 'c, maybe you could constrain it to 'a -> 'a -> 
'a.

> One more quick question.  Why did Ocaml's developers make it a point
> to separate integer computations from floating point computations?  I
> think I have an idea, but not really sure.  I know that computers can
> represent integers exactly, but when it comes to floats ( or doubles
> in other languages ) computers have to convert between base-2
> fractions and base-10 fractions, and the conversion never yields a
> value that is 100% accurate. But most languages freely go back and
> forth between ints and floats without worrying too much about any
> major loss in precision.  Why did Ocaml's developers decide to
> completely separate integer computations from floating-point value
> computations?  It's a very distinctive feature of the language that I
> have not found yet in other computer languages that I have
> experimented with.


Floats and integers are two very different types of data and 
implicitely converting one to another can have many surprising side 
effects.
At least in my experience I only rarely need float_of_int or 
int_of_float.


Christopher


-- 
http://gmerlin.de
OpenPGP: http://gmerlin.de/christopher.pub
2779 7F73 44FD 0736 B67A  C410 69EC 7922 34B4 2566