Re: Re: about Coolness, Impatience and Complexity

I recognize this module as an interesting insight into a possible evolution of computers. I do understand that it describes methods that, if implemented in hardware with the appropriate technics, would:
1. be a great step forward (precisely for the parallel computing Damian underlines it would allow);
2. imply that we consider the writing of algorithms differently from what we do now with our 'sequential' machines.

I live in a real world, work with a real PC, and have seen real people using this module. Its documentation does not say "this is a joke", and it is really available on CPAN.

If it was intended only as a joke, I had no way to know it, as its documentation doesn't say so. Moreover, for the reasons stated above, and for the fact that it provides a concise, expressive, elegant way of doing certain things (including computing minimums if you are not in a hurry) I do not think it should be considered <it>mainly</it> as a joke.

But I probably put too much focus on Q::S. I mainly intended it as an example to illustrate one of the possible shortcomings of <it>"concise, expressive, elegant"</it> ways of doing things: people misunderstanding how much calculus there is behind a simple, innocent, pretty expression.

Concerning the particular complexity of this prime number tester, I have to agree with you, Abigail. <it>It's just using different units than I am used to.</it> The problem is that the units are not specified. I did understand that Damian was expressing his complexity in terms of elementary quantic operations. But please admit that in everyday programming, what interests most of us is complexity in terms of time (or space, sometimes). This O(1) is pure prospective and does not mean much for the pragmatic programmer of today.

For your specific example of the point and the line, I agree too: I need, basically, 2 multiplications, one substraction, and one comparison. So it runs in constant time on everyday PCs (no matter how long the line is :).
Still, I can say it is an operation in O(n) (time) if I consider my little brain as the target platform (the larger the number, the longer it takes me to compute multiplications). But if I say so, I will definitely not forget to specify the unit (time) and the platform (you servant), for they are not what people are accustomed to.

(Maybe I should try and be more concise myself.) BTW, you seem to like polemics, Abigail, and I like that too :).

Blop!

Comment on Re: Re: about Coolness, Impatience and Complexity

Replies are listed 'Best First'.
Re: Re: Re: about Coolness, Impatience and Complexity by Masem (Monsignor) on Jul 12, 2001 at 22:30 UTC
Sometimes, and this is on a case by case basis, simplication, abstraction, or any other way to reduce code may be simplier to maintain, document, and explain to others than code that might run faster but is overall a bit harder to read. For example, given hypothetical: `my @d = 2 * all( @values ); # Method 1, assume 'use Q::S' my @d = map { 2$_ } @values; # Method 2` [download] (and include method 3, which would be the functional/Haskill approach to this problem, which I know you can write, but I have no idea how to write it myself :-). Method 2 is the fastest of all these, but to those not versed in perl, it may be harder to describe than method 1, which 'reads' like what it's supposed to do. Depending on the situation I was in when coding this, I might purposely using the Q::S method if only to convey meaning moreso than method 2; that situation would most likely be limited to scientific groups with understanding of computing but no strong computer language skills. On the other hand, if I was with a web site design shop, I'd automatcially advocate the use of method 2, which is straight forward from (what we hope to expect) the viewpoint of experienced computer programmers. Dr. Michael K. Neylon - mneylon-pm@masemware.com \|\| "You've left the lens cap of your mind on again, Pinky" - The Brain*	[reply] [d/l]

Replies are listed 'Best First'.

Re: Re: Re: about Coolness, Impatience and Complexity
by Masem (Monsignor) on Jul 12, 2001 at 22:30 UTC

For example, given hypothetical:

my @d = 2 * all( @values );  # Method 1, assume 'use Q::S'
my @d = map { 2*$_ } @values; # Method 2
[download]

Method 2 is the fastest of all these, but to those not versed in perl, it may be harder to describe than method 1, which 'reads' like what it's supposed to do. Depending on the situation I was in when coding this, I might purposely using the Q::S method if only to convey meaning moreso than method 2; that situation would most likely be limited to scientific groups with understanding of computing but no strong computer language skills. On the other hand, if I was with a web site design shop, I'd automatcially advocate the use of method 2, which is straight forward from (what we hope to expect) the viewpoint of experienced computer programmers.

Dr. Michael K. Neylon - mneylon-pm@masemware.com || "You've left the lens cap of your mind on again, Pinky" - The Brain

[reply]
[d/l]