Author: Heiner Marxen
Date: 04:52:01 07/15/01
Go up one level in this thread
On July 15, 2001 at 03:34:25, Vincent Diepeveen wrote: >On July 14, 2001 at 12:15:40, Heiner Marxen wrote: > >Same challenge to you, >convert the 0x88 generator from Bruce to LISP instead of >academic nonsense about how well it is. Hello Vincent, I don't consider LISP a good example of a functional language. Many years ago I have read loads of LISP code written by other people, which I expected to be LISP experts. I didn't like the results. Too many (((()))), too long, hardly readable. But there are many other functional languages. I have used with great success SML/NJ. I cannot talk about Gofer or Haskell, since I never used them. >I've written stuff in functional languages and the PRACTICAL >slowdown i measured was huge. > >One or 2 theoretic examples aren't representing truth for me. My experience was not "theoretic". I have used SML to write my "busy beaver" program, which aims to compute Rado's Sigma function. See my web page: http://www.drb.insel.de/~heiner/BB/ First I had a C program. Then I wanted a certain new feature. It was non-trivial, so I started to use C++, which I used for 3 years already on work, i.e. I was fluent in C++. I did not manage to come up with a working C++ implementation. A friend said "I would use SML", so I asked him to teach me SML. He did so in a few days. Then I implemented the complete C program and the new, wanted feature in SML. The resulting SML program was demonstrably faster than the old C program. It also had less implementation restrictions, was easier to debug and easier to extend. I am not going to convert you from a "LISP hater" into an "SML lover". Therefore I will not spend any significant time to write some move generator which I do not want/need, just to demonstrate something to you. I just tell you my experience, as you told us your experience. They differ. Well. >LISP is a laughable language, it's completely unreadable as soon >as a program gets bigger as a few pages in size, because of the >many (((((((((((((((((((((((((((((((((((((((((((((((((((((((((( >)))))))))))))))((((((((())))))))))))))(((((((())))))))))))) >(((((((((((((((((((((((((((((((())))))))))))))))))))))(((((((((((())))) >((((((((((((((((((((()))))))))))))))))))))))) >you need. Agreed, see above. If you think we should continue this topic, we probably should take this to email. Best regards, Heiner >>On July 14, 2001 at 11:53:51, Ricardo Gibert wrote: >> >>>On July 13, 2001 at 22:34:34, Dan Andersson wrote: >>> >>>>I can't say that I agree about your assertions about functional programming >>>>languages, not now anyway. Both Common Lisp and OCaml rival the speed of C. Any >>>>problems with those languages are usually the result of naive implementations. >>> >>> >>>Your assertion that they can rival the speed of C is not too surprising, since >>>both OCaml and Common Lisp also support imperative programming. Any language >>>that effectively includes C as a subset can rival C in speed. >> >>Using the imperative features of functional languages like SML tends to >>slow them down. This is mainly due to the optimization strategies of >>such compilers. >> >>Also, according to my own experience with SML compilers, truely functional >>code most of the time does compete with C or C++ in speed. >> >>Since this starts to be off topic, I will not elaborate here. >>There are several web pages which present profilings across imperative and >>functional languages, which support this experience. Just try google. >> >>Regards, >>Heiner
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.