[Caml-list] checking for same ctor type?

[Caml-list] checking for same ctor type?

[Chris Hecker]

What's the right way to check if two variants have the same ctor, but
not necessarily the same values in the ctor parms?

For example, I want to check if both data_types are Strings:

type data_type =
    String of string
  | Float of float

let a = String "foo"
let b = String "bar"
let c = Float 2.0

let ctor_equal a b = ???

ctor_equal a b = true
ctor_equal a c = false

I can think of a couple definitions of ctor_equal, one using
Hashtbl.hash_param 1 1 and one using Obj.tag (which both appear to be
doing the same thing).  Both of these seem a bit cheesy and
implementation dependent.

Is there a better way?

Chris

-------------------
Bug reports: http://caml.inria.fr/bin/caml-bugs  FAQ: http://caml.inria.fr/FAQ/
To unsubscribe, mail caml-list-request@inria.fr  Archives: http://caml.inria.fr

Re: [Caml-list] checking for same ctor type?

[Brian Rogoff]

On Wed, 27 Jun 2001, Chris Hecker wrote:
> What's the right way to check if two variants have the same ctor, but
> not necessarily the same values in the ctor parms?
> 
> For example, I want to check if both data_types are Strings:
> 
> type data_type =
>     String of string
>   | Float of float
> 
> let a = String "foo"
> let b = String "bar"
> let c = Float 2.0
> 
> let ctor_equal a b = ???

let ctor_equal a b = 
  match a,b with 
  String(_),String(_) | Float(_),Float(_) -> true | _,_ -> false

> ctor_equal a b = true
> ctor_equal a c = false
> 
> I can think of a couple definitions of ctor_equal, one using
> Hashtbl.hash_param 1 1 and one using Obj.tag (which both appear to be
> doing the same thing).  Both of these seem a bit cheesy and
> implementation dependent.

Yes, but there's only one implementation, right? 

> Is there a better way?

Beauty is in the eye of the beholder. 

-- Brian


-------------------
Bug reports: http://caml.inria.fr/bin/caml-bugs  FAQ: http://caml.inria.fr/FAQ/
To unsubscribe, mail caml-list-request@inria.fr  Archives: http://caml.inria.fr

Re: [Caml-list] checking for same ctor type?

[Chris Hecker]

>let ctor_equal a b = 
>  match a,b with 
>  String(_),String(_) | Float(_),Float(_) -> true | _,_ -> false

Right, I should have added "without writing out all the cases".  :)

> Yes, but there's only one implementation, right? 

Unless they never update it again, then "implementation dependent" is usually a bad thing, in my book (and the ocaml folks shouldn't want people depending on implementation details either, since it makes their lives miserable when it comes time to update stuff).

Chris

-------------------
Bug reports: http://caml.inria.fr/bin/caml-bugs  FAQ: http://caml.inria.fr/FAQ/
To unsubscribe, mail caml-list-request@inria.fr  Archives: http://caml.inria.fr

Re: [Caml-list] checking for same ctor type?

[Chris Quinn]

Pervasives.compare is fine for nullary constructors or where you don't mind the performance hit of the function recursing into the structure of the arguments.

A more efficient equality predicate that is really just a cut down version of [compare]:

module Kludge =
  struct
    let cmp (a: 't) (b: 't) =
      let a' = Obj.repr a and b' = Obj.repr b in
      match Obj.is_block a', Obj.is_block b' with
      | true,true ->
          Obj.tag a' = Obj.tag b'
      | false,false ->
          a' = b'
      | _ -> false
  end

It of course relies on the physical representation of sum types.
It has the merit of working on a type of any number of constructors, but the mis-merit of also operating over functions for instance!
To constrain such a comparison to sum types would require it to be built into the compiler.

Chris Q.
-------------------
Bug reports: http://caml.inria.fr/bin/caml-bugs  FAQ: http://caml.inria.fr/FAQ/
To unsubscribe, mail caml-list-request@inria.fr  Archives: http://caml.inria.fr