English version
Accueil     À propos     Téléchargement     Ressources     Contactez-nous    

Ce site est rarement mis à jour. Pour les informations les plus récentes, rendez-vous sur le nouveau site OCaml à l'adresse ocaml.org.

Browse thread
Pervasives.compare output type
[ Home ] [ Index: by date | by threads ]
[ Search: ]

[ Message by date: previous | next ] [ Message in thread: previous | next ] [ Thread: previous | next ]
Date: 2005-03-30 (17:47)
From: brogoff <brogoff@s...>
Subject: Re: [Caml-list] Pervasives.compare output type
On Wed, 30 Mar 2005, Jacques Carette wrote:

> Alex Baretta <alex@barettadeit.com> wrote:
> > Xavier Leroy wrote:
> > > It's a historical error.  [...]
> >
> > Whether fixing such historical errors engenders more benefits than trouble is a very interesting philosophical
> > question.
> It all depends on whether the installed base is viewed as more/less important
> as the future integrity of the software system as a whole.  When I was in
> industry, I was in a position where I made the choice that future integrity
> was more important [and I did annoy the user base but improved the basic
> system a lot], and my successors made the opposite choice [which means that
> mostly do 'new' features and can't fix some well-known bugs that users have
> learned to work around].

Speaking as an industrial user, I don't see it as such an either/or. I'd
prefer that enhancements (thisisn't a bug) which require rewriting code
be done as few times as possible, so that each release doesn't require
rewriting code. So, let's say for example that the implementors decide to
change this, and to fix evaluation order to be left to right, I'd prefer
that it were done in one release, rather than two separate ones.

I think even industrial users should realize that OCaml is a research langauge,
and while we're grateful that it is generally quite stable from relase to
release, that research goals take some precedence. Maybe one day the
implementors will decide to "radically" change things, as in the move from
Caml Light to Caml Special Light/OCaml.

-- Brian