[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: exception systems

From: kelsey@research.nj.nec.com (Richard Kelsey)
Subject: Re: exception systems
Date: Thu, 18 Apr 1996 19:02:55 -0400

>    References: <199604181941.PAA15530@sting14>
>    Date: Thu, 18 Apr 1996 18:15:59 -0400
>    From: Matthias Blume <blume@CS.Princeton.EDU>
>    I still think there is way too much in it.  Where is MI suddenly
>    coming from?  If I recall correctly, we don't even have SI in Scheme
>    -- and I'm very happy about this fact.
> So am I.  My initial reaction to multiple-inheritance was Uuugh!
> I changed my mind when I was unable to come up with an reasonable
> alternative.  Note that I am not suggesting adding multiple
> inheritance for anything other than conditions.

But shouldn't this prompt us to ask why we seem to need this `feature'
at all.  ``Programming languages should be designed...''

> I think you are missing the point.  The issue isn't functionality.
> As I said in the proposal it could all be implemented easily in
> Scheme.  The issue is interfaces.  The notion is that you should
> be able to say for procedure <foo>:
>  If <foo>'s argument is not a <bar> it installs an <interactive>
>  restart that accepts a new argument and then signals a <wrong-
>  type-argument> condition.

`wrong-type-argument' shouldn't be a run-time issue at all.  But this
has been discussed and disagreed-on before.

>    I would be perfectly happy with something as simple as
>    `raise'/`handle'.
> Which is exactly what I proposed originally.  The consensus
> was that it was completely inadequate.  You didn't comment at
> the time.

My mistake.


I have realized before that Scheme is not moving in a direction that I
can agree with.  This whole exceptions thing is just another instance
of this phenomenon.  When I came to Scheme I liked it for its
simplicity and clarity, which explains why I now dislike it more and

If I were interested in a Common Lisp, I'd know where to find it.