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

Re: Voting with our wallets

Brad Miller writes:

> Guys, there's a market for this. We aren't buying many more Sun
> workstations, and the number of Macs at our site has been
> growing. Virtually every prof now has a powerbook, and technical
> staffers have macs at home. A good many profs got exposed to lisp for
> the first time by way of MCL.

> If you're gonna write it off, at least include it with every mac sold
> for a while. That's how hypercard got started. It's also how Microsoft
> Basic got started.

Brad has inspired me to get something off my chest that's been bugging 
me for some time. Apple isn't actively working on the much-needed
PowerMac port because all their efforts are going into Dylan. This is
an amazingly naive and unrealistic decision on Apple's part. Why?
Well, even though Dylan may incorporate some really good design
ideas, how much credibility is it going to have in the real world?
Dylan might get used on some isolated development projects on the
Mac and maybe Newton-like devices produced by Apple, but it will
never be considered a serious language on other platforms. Apple-
developed software has exactly zero serious cross-platform
credibility, even though it's usually superior. Don't believe me?
Well, how about DAL, OpenDoc, and AppleEvents, for example? Microshaft's
vastly inferior "me too" versions of these tools are the only
seriously-considered products in the PC world. TrueType is the one
exception, and sadly, it's also the only Apple product that was a
serious goof from a technical standpoint.

On the other hand, Common Lisp/CLOS is a recognized cross-platform
standard and is already supported by multiple vendors. It may not
be as blindly-accepted as silly junk like C++, but at least it's
considered "real". Dylan will never be much more than an interesting
academic curiosity when it comes to developing real-world applications.

And now the Dylan team has bastardized its syntax in a move to lure
the C sheep, but all they've really done is alienated those of us who
know infix notation is a bad idea.
We have Common Lisp running as Symbolics Genera on an XL1200, Genera
on a DEC Alpha, Harlequin on the DEC Alpha, and Franz on a Sparc 10.
MCL execution speed on a Quadra 800 beats them all. BEATS THEM ALL!
This is obviously a testament to just how good MCL's compiler is
(and how poor the others are). If this thing ran native on a PowerMac 
it would be the most amazing Lisp product ever invented.

If Apple is as serious about dynamic object-oriented languages as
they say they are, they should reverse their priorities and start
working full-speed on MCL and play with Dylan in their spare time.

Flame off.
Shannon Spires
Internet: svspire@sandia.gov
Voice: 505-844-4287
Fax: 505-844-3296