Re: 3.99.3

From: Russ McManus <>
Date: 06 Oct 1998 09:11:54 -0400

Christian Lynbech <> writes:

> The different implementations are too far apart in their goals and
> architectures. A true merge will require the different camps to
> compromise and that is simply not going to happen.

This sounds like a self fulfilling prophecy.

> Just try to imagine the scheme crowd trying to agree on what kind of
> OO system such a common implementation should support (and no, just
> throwing them all in isn't going to make things better). Don't forget,
> the fathers of R4RS left out macros because there was no perfect
> solution to the numerous limitations of the old-style defmacro (which
> of course led to each implementation having its own variation on the
> theme).

Why can't multiple OO systems be built on top of the same scheme
system? It won't make OO programming in scheme any better, I agree,
but it will reduce the duplication of effort in maintaining multiple
implementations. I think it is still a win.

> Even merging STk and Guile will be difficult, even if they have the
> same ancestor. Could be that STk is a leaner and meaner implementation
> than Guile, but I think that Guile rocks in some other such as
> debugging and posix interfacing. If we moved all that to STk, would
> STk then still be that much leaner and meaner?

By applying your reasoning, one can deduce that STk would cease to be
lean and mean. In the case you cite, that would be a cost of merging
implementations. I agree that there are costs; I just think that the
benefits greatly outweigh the costs.

> IMHO what we need is more a broader standard such that programs
> become more portable, rather than a single unified
> implementation. CLTL2 may be big but one can sure get a lot longer
> in pure CLTL2 than in pure R4RS. I also think it will happen. STk
> has shown the way in CLOS style OO, it will make its way into guile,
> perhaps into others. Module systems are being worked on, eventually
> one will dominate. FFI is being worked on etc. In time, we will
> overcome the problems. Lisp is not as big as java, probably never
> will be, but it isn't dead.

I disagree about your theory that eventually the best Scheme features
will dominate, if only because eventually our planet will be swallowed
by the sun. ;^)

The notion of bringing together scheme implementations through
expanded standards has merit, though. Too bad there has not been any
move towards an expanded scheme standard in 25 years.


"NT 5.0 is the last nail in the Unix coffin. Interestingly, Unix isn't
in the coffin... It's wondering what the heck is sealing itself into a
wooden box 6 feet underground..." 
             -- Jason McMullan
Received on Tue Oct 06 1998 - 15:12:41 CEST

This archive was generated by hypermail 2.3.0 : Mon Jul 21 2014 - 19:38:59 CEST