Author: Vincent Diepeveen
Date: 09:53:17 01/10/03
Go up one level in this thread
On January 10, 2003 at 12:28:56, Uri Blass wrote: It was written in emacs. It is doing things like { } that's 3 lines extra for free. #include <stdio.h> #include <string.h> bla bla. all default crap. Ask people how fast i can type :) 400 keys a minute. all default crap. then some cut'n pasting of parsing stuff and changing 2 letters in it and 1 resulting action and you get lines for *free* :) the resulting thing was somewhere against 200 lines. It's an estimation from now :) the total work was 30 minutes in total and that included 1 bathroom break. I do not understand why they estimate effective work done in businesses at 0.5 lines an hour. That means i work 200 times more efficient when i program computerchess features (not to confuse with algorithms or optimizing the code a lot). features is straight programming. Of course you never manage that 8 hours a day. I remember my record was some months ago when i made an entire new EGTB scheme from scratch with new features and such. My old EGTB scheme was very inefficient. The straight feature programming was done in a matter of a day. No cut'n pasting at all. All made from scratch with perhaps a few lines exception (initializing of KK reduction which i did in old EGTB). Rest all written from scratch in a day. 10/23/2002 05:14p 56,776 DIEPBITT.C The bugfixing took however quite some longer :) Even faster than this straight ansi-coding goes coding interfaces. it's cut'n pasting so many lines for each new feature and then modifying a few things that those standards of 0.5 line an hour a programmer is IMHO completely outdated. >On January 10, 2003 at 10:52:37, Vincent Diepeveen wrote: > >>On January 10, 2003 at 08:01:45, Dan Andersson wrote: >> >>>A couple of postings on comp.lang.scheme implemented the CoyoteGulch benchmarks >>>in Scheme. And lo and behold BigLoo produced faster code for the numerical >>>benchmark. One might object that since BigLoo emits C or Java bytecode it isn't >>>really faster. But the amount of automated program transformations that are >>>applied is huge. For a coder to do the same thing would be like trying to >>>outperform a spreadsheet. And the C code is inhuman in nature. And the question >>>arises: Why on earth would one use C++ or Java? Both are verbose and terribly >>>low level compared to lambda calculus. >>> >>>MvH Dan Andersson >> >>I disagree. >> >>When i converted my functional international checkers program to C it was not >>bigger in code and it was 10000 times faster in C than in Gofer (which also has >>lambda). >> >>Later efforts from expert gofer programmers speeded it up another 2 times. >> >>Big deal. factor 5000 difference. >> >>Compiling it was not very well possible that gofer language, because there were >>no good compilers for it (the only thing that was there was some stupid turbo-C >>thing and by then turbo-C was completely outdated and only worked in DOS not >>windows). >> >>However on a fast PC i tried and the difference after some tuning was still a >>factor 100 favourable for C. >> >>It is not true that functional programming is easier. Instead it is more >>difficult to learn because you need more functions and shorter functions. The >>lambda you will never manage to explain whereas the imperative languages are >>clear for everyone. >> >>The big 'advantage' of gofer was said to be easy parsing anything. Then they had >>a script to 'parse' in such a way that it could be used to use gofer for CGI >>parsing. >> >>Then i had a contest between someone who promoted on Gofer (so not exactly a >>gofer beginner) who would be using gofer mixed with haskell to parse some type >>of output from a homepage into a textfile and then converting it back to a >>resulting html page which showed results. >> >>I would write it in C. He in gofer+haskell (whatever). >> >>After 10 minutes i was nearly done and after going to the toilet and then fixing >>another bug after in total 30 minutes i finished my C parser and had it bugfree >>working and doing anything and everything. >> >>After 2 hours he still didn't manage to write the equivalent in Gofer+Haskell, >>even though he could use a 'cool' number of already existing scripts whereas i >>parsed the direct (method POST) output. >> >>My CGI script in C was like 200 lines at most but it easy to write and easy to >>debug. It took another few afternoons before the equivalent in Gofer was ready. >> >>30 minutes including bathroom stop in C is equivalent to some hours of >>functional coding. >> >>Best regards, >>Vincent > >I am surprised that you can write 200 C lines in 10 minutes with only one bug. >ut means that you write one line in one second. > >To be more correct you say at most 200 lines but even if I assume that it was >only 100 lines still writing one line in 6 seconds seems impossible for me. >Only writing your program from memory (if I remember all the program) seems to >me more time. > >I also almost never write 200 lines without some testings in the middle for >bugs. > >Uri
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.