are initial values in class fields mandatory ? (ocaml)

are initial values in class fields mandatory ? (ocaml)

[fbesnard]

I'd like to define a class in Ocaml without giving intial values to its fields. Is it possible ?

For example, as far as I've seen, a class with a int and an int list fields would look like this:

class dummy () =
	val mutable v = 0
	val mutable l = ([] : int list)
end;;


but what about a field of type t, for which there is no default value ?
i.e.

	val mutable v = ( ??? : t )


other questions:
- are documentations available in pdf (Acrobat Reader) format ?
- there is no 'per line' comment (like Ada's "--" or C++'s "//" ) ? 

Re: are initial values in class fields mandatory ? (ocaml)

[Donald Syme]

> - there is no 'per line' comment (like Ada's "--" or C++'s "//" ) ? 

I too think this is a must.  It may not quite accord with the personal 
tastes of the CaML implementors, but many good programmers use this
extensively in their in-line documentation.  I figure they should be 
able to transfer their "existing practice" when they make the switch to 
CaML, just to make the switch as painless as possible.

Cheers,
Don Syme


-- 
-----------------------------------------------------------------------------
At the lab:                                          At home:
The Computer Laboratory                              Trinity Hall
New Museums Site                                     CB2 1TJ
University of Cambridge                              Cambridge, UK
CB2 3QG, Cambridge, UK
Ph: +44 (0) 1223 334688                              Ph: +44 (0) 1223 563783
http://www.cl.cam.ac.uk/users/drs1004/
email: drs1004@cl.cam.ac.uk
-----------------------------------------------------------------------------

Re: are initial values in class fields mandatory ? (ocaml)

[Francois Pessaux]

Hello,

> I'd like to define a class in Ocaml without giving intial values to its fields. Is it possible ?
No

> For example, as far as I've seen, a class with a int and an int list fields would look like this:
> 
> class dummy () =
> 	val mutable v = 0
> 	val mutable l = ([] : int list)
> end;;
> 
> 
> but what about a field of type t, for which there is no default value ?
> i.e.
> 
> 	val mutable v = ( ??? : t )
> 
That the same problem than trying to create an int ref with no default value.
You can't and you don't want !

> other questions:
> - are documentations available in pdf (Acrobat Reader) format ?
No

> - there is no 'per line' comment (like Ada's "--" or C++'s "//" ) ? 
No

-- 
(*                      Francois PESSAUX (Francois.Pessaux@inria.fr) *)
(*                               INRIA Rocquencourt - Projet CRISTAL *)
(*                               (http://pauillac.inria.fr/~pessaux) *)
;;

Re: per-line comments

[Xavier Leroy]

> > - there is no 'per line' comment (like Ada's "--" or C++'s "//" ) ? 
> I too think this is a must.  It may not quite accord with the personal 
> tastes of the CaML implementors,
  
Caml's comment syntax is more a question of historical tradition than
personal tastes.  After all, the comment syntax is perhaps the only
bit of syntax that hasn't changed between LCF ML, Caml, and Standard ML...

> but many good programmers use this
> extensively in their in-line documentation.

I don't quite get the point about in-line documentation.  I write
perfectly fine on-line documentation between (* and *).  Why would it
be any better with // or -- ?

> I figure they should be 
> able to transfer their "existing practice" when they make the switch to 
> CaML, just to make the switch as painless as possible.

Depends from which background you come.  E.g. for a well-trained C
programmer, delimited comments make more sense.

At any rate, the main problem with per-line comments (as with most
syntax extension) is that they need a reserved symbol.  Both // and --
are valid OCaml infix identifiers; // is even used in the Num standard
library module.

- Xavier Leroy

Re: per-line comments

[Pierre Weis]

[French abstract ahead]
> Caml's comment syntax is more a question of historical tradition than
> personal tastes.  After all, the comment syntax is perhaps the only
> bit of syntax that hasn't changed between LCF ML, Caml, and Standard ML...

Even this bit of syntax has changed: LCF ML comments where % blabla %

> > but many good programmers use this
> > extensively in their in-line documentation.
> 
> I don't quite get the point about in-line documentation.  I write
> perfectly fine on-line documentation between (* and *).  Why would it
> be any better with // or -- ?

Right. But many programmers feel it simpler to end the comment with a
carriage return (if not for the historical trick of being able to put
a new instruction after the comment and to reuse the same card upface
down!  This is a dinosaure's story from the good old time of real
computers that knew the meaning of a column :)

> At any rate, the main problem with per-line comments (as with most
> syntax extension) is that they need a reserved symbol.  Both // and --
> are valid OCaml infix identifiers; // is even used in the Num standard
> library module.

Perfectly exact and pertinent. Even the (* delimitor is a problem when
you want to name operators, since (+), (-), (/) work perfectly but you
must use extra spaces for the multiplication.

However, the old meaning of % in old LCF ML, was carried to Caml, and
may be for this historical reason the % character is not in the
list of operators for Caml Light or O'Caml. Thus it is usable as a
comment delimitor, if not more useful for another purpose.
In addition, per-line comments can be a simple way to have comments that are
``free style'' comments, as opposed to the actual comments that must
be a suite of lexically correct tokens.

Thus it is possible to add per-line comments, but Xavier pointed out
the main problem: is it desirable to add this feature to the langage?

[Résumé en Français]
Xavier a raison: -- et // sont déjà pris. D'ailleurs (* pose déjà un
problème avec la syntaxe des opérateurs infixes (il faut écrire ( * )
pour la multiplication alors que ces blancs sont non nécessaires pour
les autres opérateurs). Ceci dit % est inutilisé (et il a toujours
signifié début de commentaire depuis le début de ML et même en Caml
jusqu'à 1991) et pourrait servir à faire des commentaires uni-lignes
(et à l'occasion des commentaires qui ne sont pas une suite de lexèmes
valides du langage). Mais est-ce vraiment nécessaire ?

Pierre Weis

INRIA, Projet Cristal, Pierre.Weis@inria.fr, http://pauillac.inria.fr/~weis/