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

Re: Allegro Composer vs Symbolics

It's been buggy for the better part of a year - how much longer
until the bugs go away? Try making a huge cons'd together structure
(that is, start eating up all of memory) - like this:

(defun cons-a-lot (n-levels)
  (make-array 10
	      (make-list 10
			 :initial-element (if (zerop n-levels)
					    (cons-a-lot (1- n-levels))))))
    the above code due to Charles Fry (chucko) of Symbolics

try increasing orders of magnitude, beginning with 10 (try this under
genera, too). I found as of last fall ('89), neither Lucid nor Allegro
(I had what was supposed to be the latest version of both) could cleanly
deal with running out of space. Lucid just disappeared, and Allegro
did different things - most commonly it would say it was running out of
stack space, should I increase it, and if I said yes, it would crash ...
back to a login (unix login, mind you) prompt. We were also looking at
Harlequin, and it had similar unfortunate behaviour, but I can't remember
what it was. Now, I never ran the Symbolics up against its hard memory
limit, but I went two *orders of magnitude* higher with the symbolics
and it still came back to me. Yes, the symbolics had to grow the stack
several times, but it did that successfully, and it garbage collected
the lot successfully.

Now, the above is a LISP problem - Composer, Harlequin, and SPE had
bugs, too.

BUT, I agree with one point you're making. When one looks at the
functionality that Composer et. al. deliver, and compare such a list
with genera, genera doesn't stack up as well as it might. Now, most
of us know that the difference in robustness, etc. is substantial,
but when the bottom line is as different in price ('cause Allegro
Composer, etc run on the workstations we already have hundreds of,
so we're talking 4K - 5K as compared to 23K to turn a sun into a
lisp workstation) how do you go to the people who are paying the
bills (or better, for how much longer are you going to be able to
go to the people paying the bills) and be able to justify spending
4 to 5 times the amount of money for no discernable (on paper)

(a) Symbolics marketing should do an aggressive comparison of their
development environment's features vs. the competition and at least
post same to the net, if not distribute it to the sales force.

(b) Symbolics should be considering how to keep Genera out in front
in the future - one idea involving the "bundling of Concordia and
Statice" discussion that's slowly disappeared. Maybe Symbolics will
look at some of those ideas a second time when we see what they
come up with for (a), instead of telling us that all we're looking
for is price cuts.

Want a chilling thought? Say you wanted to buy a Symbolics and
some state institution you work for insists that it go out for
bid. How much difference do you think some dweeb in a gov't
purchasing office is going to see between all of the aforementioned
lisp development environments? Probably about $15K. For now, I
can make robustness arguments that might win, and (fortuanately)
I can make compatability arguments that do win. Its not clear to
me that functionality is going to remain a convincing argument
for long.