[plt-scheme] xxx chooses MzScheme as preferred language

From: Matthias Felleisen (matthias at ccs.neu.edu)
Date: Sat Jun 3 08:40:51 EDT 2006

On Jun 3, 2006, at 12:48 AM, Abdulaziz Ghuloum wrote:

> Yes, Schumacher builds fast cars unsuitable for use as a family sedan 
> and
> a Lexus does not belong in a Formula 1 world championship.  It's also 
> true
> that Chez Panisse (pun not intended) and McDonald's are not really in 
> the
> same business.  But back to Scheme implementations ...

To compare PLT Scheme to MacDonalds is ludicrous. The very moment
you choose Scheme, you're already in the business of high end food.
It just so happens that some engineer food for the billionaires
who once a year eat live frogs and camel backs; and the other engineers
high-end nouveau cuisine that is currently hip with the young.

> I look at PLT's work and I admire how much effort was put in the 
> frontend,
> the user interface, and the whole experience (modulo the speed thingy).

Front end? Have you ever read the OS paper from ICFP Paris?

> I look at Chez and just can't get over the effort that was put in 
> making
> it the finest implementation (modulo the missing human interface 
> thingy).

I agree hands down that Chez is the best engineered Scheme
implementation in the world. Bob H (when he was alive) and
I used to talk a lot about Kent's practices and I have nothing
but respect for this work.

> Can't there be an implementation that is both at the same time?  I 
> mean,
> there is no magic in making good human interfaces as DrScheme already
> demonstrates.

In addition to the human interface, there is lots more. You just don't
see it.

> Making fast optimizing compilers is not magical either
> since every single aspect of Chez Scheme (its runtime architecture, the
> optimizations it performs, its register allocator, its garbage 
> collector,
> etc.) has been described in details in the literature.

In addition to the published research results in Chez, there is lots 
You just don't see it.

Both are extremely expensive projects with N manyears of work behind
them for very large N. In Kent's case, there is a ton of industry money
in it. In PLT's case, there is a ton of NSF and Edu and donation money
in it. Not to speak of the sheer love that went into the construction
of both. Yes, I really meant "love".

Now how do you combine two things of that order of magnitude? It would
be a hyper-expensive proposition.

At this point, you need to ask yourself why the Scheme community is so
splintered. And why it remains so splintered. I am not a sociologist but
I think all of us can analyze and formulate conjectures and thrown them
away before they create more havoc.

-- Matthias

Posted on the users mailing list.