Why we should remove {} from Squeak

Richard A. O'Keefe ok at atlas.otago.ac.nz
Wed Oct 3 04:23:35 UTC 2001


danielv at netvision.net.il wrote
in response to my question "what's wrong with reductio ad absurdum
arguments":
	So they're a long way of saying "what you care about, I'm ignoring",
	which is argumentative and not fun.
	
A reductio ad absurdum argument proceeds by showing that the alleged
grounds for some proposal lead to a conclusion that the proposer is
expected to dislike.  As such, it is precisely *NOT* ignoring what the
proposer cares about, to think that is to profoundly misunderstand their
nature and force.

As a concrete example, if I seriously supposed that Stephane Ducasse
*would* like an otherwise Smalltalk-like language containing no literals
other than numbers and no blocks, then the argument "if you push hard
on 'simplicity' and 'purity' you get this" would be pointless.
'Simplicity' and 'purity' weren't _my_ concerns, they were _his_;
my attempt to persuade him that {} should not be removed from Squeak
would be futile if I didn't suppose that _he_ is happy with blocks and
string literals.

	By another measure, they very often don't include any new ideas or
	interesting presentation, which is a waste of bits.

That's a curious definition of "a waste of bits".  Suppose I had been
able to persuade the damned (and I mean that quite literally) terrorists
of recent terrible memory not to commit their evil deed by e-mailing them
selected passages from the Koran (which as a matter of fact I have checked
to see if it says anything that bears on this issue, and it does, it does,
all negative; NO WAY could that be justified from the Koran), then my
hypothetical e-mail would not have contained any new ideas, nor any
particularly interesting presentation (I mean, these people are supposed
to _know_ the Koran), yet I do not think it would have been a waste of bits.
[This is another reductio ad absurdum argument.]

Sometimes we need to be reminded of things we already know.

By the way, the fact that I do not regard the arguments we have seen so
far as sufficient reason to remove {} does NOT mean that no such arguments
could exist, nor even that I do not think that such arguments might exist.
A reductio does not show that the conclusion is untrue, only that the
supporting argument is weak.

If Smalltalk had started with {...}, I wonder whether #(...) would ever have
been invented?  I wrote a Smalltalk parser in C once, to investigate some
ideas about compression.  #(...) was one of the three trickiest things I
had to deal with, {...} was no trouble at all.





More information about the Squeak-dev mailing list