From mboxrd@z Thu Jan  1 00:00:00 1970
Received: by pauillac.inria.fr; Wed, 2 Mar 94 15:09:00 +0100
Received: from margaux.inria.fr by pauillac.inria.fr; Tue, 1 Mar 94 18:55:07 +0100
Received: from pauillac.inria.fr by margaux.inria.fr, Tue, 1 Mar 94 18:55:04 +0100
Received: by pauillac.inria.fr; Tue, 1 Mar 94 18:55:03 +0100
Date: Tue, 1 Mar 94 18:55:03 +0100
Message-Id: <9403011755.AA26981@pauillac.inria.fr>
From: Valerie.Menissier@inria.fr (Vale'rie Me'nissier-Morain)
Sender: weis@pauillac.inria.fr
Organization: INRIA - BP 105 - F-78153 Rocquencourt Cedex
              Tel.: (33 1) (16 1) 39 63 55 98 Fax: (33 1) (16 1) 39 63 53 30
To: caml-list@margaux.inria.fr
In-Reply-To: Francisco Valverde Albacete's message of Wed, 16 Feb 94 10:54:22 +0100 <9402160954.AA18004@lorien.die.upm.es>
Subject: Some questions and proposals...

>	2) This is a question mainly to CAML and Camllight implementors: has
>	anybody thought of a way to migrate in an automatic, or at least semi-
>	automatic way, modules from one of the systems to the other? I
>	have this huge quantity of modules in CAML wanting to be
>	compiled in Camllight...!

In fact the difficulties to port Caml programs to Caml-Light seem to
me of three orders:

1. 
   a. Functions written in the prelude file of Caml that are not
defined by Caml-Light: this is easy to correct, just try to compile
and for each unbound name, go in Caml prelude and copy-paste the
definition in a compatibility file loaded before your programs, for
example 
  b. Functions that are curryfied in only one version, generates a
typical typechecking error, so it seems easy to change
  c. Functions with arguments in another order, either generates a
typechecking error and it is easy to change, or for example functions
like "move src dest" changed in "move dest src", and it seems very
difficult to detect 
  d. Same function, same arguments, but not same result, for example 
"index" function for which in one version the first character is
number 1 and in the other version it is number 0

2. Functions that use the fact that the order of evaluation of Caml is
from left to right: it seems very difficult to automatize this phase

3. Functionalities that are different in Caml and Caml-Light like
parsers&printers vs. grammars or some differences in arithmetic, it
seems very difficult to automatize this phase

Vale'rie Me'nissier-Morain