Re: The option -rectypes

From: Jerome Vouillon (Jerome.Vouillon@inria.fr)
Date: Thu Nov 25 1999 - 23:44:22 MET


Date: Thu, 25 Nov 1999 23:44:22 +0100
From: Jerome Vouillon <Jerome.Vouillon@inria.fr>
To: Gerd.Stolpmann@darmstadt.netsurf.de, caml-list@inria.fr
Subject: Re: The option -rectypes
In-Reply-To: <99112402223302.32002@ice>; from Gerd Stolpmann on Wed, Nov 24, 1999 at 01:59:36AM +0100

On Wed, Nov 24, 1999 at 01:59:36AM +0100, Gerd Stolpmann wrote:
> I have some code that compiled in 2.02, but in 2.03 I need the new option
> -rectypes of the compilers.

This is a bug in the occur check. It will be corrected in the next
release.

> - I have thought about the type 'a node extension as 'a. If I apply the
> constraint of "node", the condition must hold that
> ('a node extension as 'a) node extension
> unifies with
> ('a node extension as 'a) node #extension
> Normally, a closed class type does not unify with an open class type,
> and I wonder why this is accepted at all.

The type of self cannot be unified with closed object type: this would
prevent a further extension of the class. But a type such as
#extension can be unified with a closed object type. In particular,
it can always be unified with the same type without "#".

-- Jérôme



This archive was generated by hypermail 2b29 : Sun Jan 02 2000 - 11:58:29 MET