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

Re: C vs. Lisp, or UNIX vs. Genera, or just US vs. THEM. Cuckoo?

	[ historical merits of lisp machines deleted ]

     My question to you is, in your move away from the Lisp Machine
     (explained by your problems with such things as your experiences with
     i/o speed, garbage collection and switching dynamic windows, as well as
     your strong attraction to Unix/C) do you wish to have the Lisp Machine

     I suspect not.

     Sadly however, I fear that the accumulation of such volumes of
     denigration (of which yours is just a tiny part) may help bring about
     that obliteration.

No, I disagree.  My goal is not to obliterate the lisp machine.  My goal is
to make my grievances about the machine known to a community that can do
something about it.

     I see two reasons for this. Firstly the mud slingings are not
     arbitrated, there is never an evaluation of the truth by a higher and
     public judge (for instance there will never be any "proving", to all
     parties, of the validity of your argument that Lisp has to create

Sorry, I disagree again.  There is a public judge.  It's called the buying
public.  "Proving" is done by measuring performance against other platforms
and advertising the results to those who might care.  Someone called for
performance comparison between lisp and C applications.  Let's do it.
If Symbolics thinks it can win (maybe it can), it's ultimately up to Symbolics
to bolster it's own public image.  SLUG readers can do it too, but Symbolics
has the ultimate responsibility.

     And secondly, in general people working in the Lisp Machine world seem
     not promote themselves (for instance, Flavors weren't touted by us, the
     users, from pillar to post throughout the first part of the last decade
     (as are more recent efforts to provide alternative tools for OOP)).

Again, marketing is not the job of the user.  It's the job of the seller.
If Symbolics had a clue about marketing it's product, I bet faithful users
would fall in line with high praise for the features of the machine.  We
should not be expected to initiate promotion, only to confirm it.

     The last Lisp Machine source is Symbolics. I for one do not want to
     waste time casting aspersions at Unix/C, but I do passionately want the
     Genera environment and the brilliant specialist systems on which it runs
     to survive. I want them to be available so that much work begun and
     other ideas as yet unformed can grow and shine.

     I don't know the figures, but the Unix/C market looks like $billions and
     the Lispm is but a tiny mote.

     Do the Unix/C people have to behave like cuckoos - eradicating
     everything else, suffering no alternatives? What in the Lispm threatens
     them - it is certainly not commercial pressure?

If the concensus is that I am acting like a cuckoo, I will cease and desist
and never post again.  It is Symbolics that I see as the cuckoos.  I am
trying to point out significant performance problems that affect my ability
to use the Lisp machine (and lisp, but I have been corrected on that point).
Two years ago, I complained about I/O.  No one refuted my complaints, some
even agreed, but after 4 new hardware and software releases, these problems
haven't even been addressed. Out of this discussion, I have been told about
CLIM.  Sounds like a step in the right direction.  But why don't the cuckoos
at Symbolics stop building new machines and fix the old ones.  Concentrate
on redesigning the features that cause the user problems.  New hardware can't
solve the problems of inefficient software.  It can only postpone them.

If, in fact,  there are no significant performance problems, test yourself
against UNIX/C and report the results.  If there are problems, FIX THEM.  You
can build XL-10million and add billions of new features, but if you don't fix
the old problems, or at least respond to the complaints of the users with proof
that the complaints are invalid.

I'm not trying to obliterate the lisp machine.  I just want the best
environment for me to work in.  It seems to me, Symbolics is committing suicide
with its own development plans.  The original developers produced a true work
of art.  But programming is not only about art, it's also about performance.
And when it comes to performance, it's either put up or shut up.

-- David Magerman
Stanford University