Author: Robert Hyatt
Date: 08:06:27 08/24/04
Go up one level in this thread
On August 24, 2004 at 09:53:20, Tony Werten wrote:
>On August 24, 2004 at 09:10:03, Uri Blass wrote:
>
>>On August 24, 2004 at 02:57:41, Uri Blass wrote:
>>
>>>On August 23, 2004 at 23:45:00, Robert Hyatt wrote:
>>>
>>>>On August 23, 2004 at 20:29:15, Gerd Isenberg wrote:
>>>>
>>>>>On August 23, 2004 at 17:41:38, Matthias Gemuh wrote:
>>>>>
>>>>>>
>>>>>>
>>>>>>>
>>>>>>>Such "standard" code like NextMove is IMHO not sufficent to proof El Chinito as
>>>>>>>Crafty clone. Did i missed something?
>>>>>>>
>>>>>>>BTW. Is it legal to disassembly others executables?
>>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>Hi Gerd,
>>>>>>Is "such standard code like NextMove" also implemented the same way in your
>>>>>>engine, IsiChess ? Are the Crafty evaluation functions in ElChinito also
>>>>>>standard and in IsiChess ? Are the reported Crafty bugs in ElChinito also
>>>>>>standard and present in IsiChess ?
>>>>>
>>>>>That IsiChess has a complete other movegen does not proof, that El Chinitos
>>>>>getMove is a copy of Crafty, even if it is likely, due to the other point with
>>>>>eval and the 99999 compare.
>>>>>
>>>>>I admit that i copied some source code here from CCC via clipboard into my
>>>>>program, for instance kogge-stone algorithms. The same might be true for others,
>>>>>and the intention i and others post source-code here is to share it and to get a
>>>>>feedback, improvements and other ideas.
>>>>>
>>>>>Is Crafty's getMove-code really so unique and some code snippets got never
>>>>>posted here? If i implement some quicksort from some published pseudo code, it
>>>>>is not unlikely that i get the same assembly, despite other identifiers.
>>>>
>>>>That's a poor argument, often tried on me by students. It doesn't fly. A
>>>>bubblesort or quicksort or heapsort written by two different people might look
>>>>the same for bubblesort (10 lines of code) but _definitely_ not for quick/heap
>>>>sort. NextMove() is over 250 lines of code. The chances of two people writing
>>>>two programs independently, and having them produce the _same_ assembly, is so
>>>>close to zero that IEEE FP would store it as zero.
>>>
>>>I agree that having the same NextMove() is wrong but I think that you do not
>>>expect people never to read the source of Crafty and learn from it so even if
>>>people do not copy Crafty they cannot claim that they wrote their programs
>>>independently.
>>>
>>>I guess that there are also small functions that a lot of people copy like
>>>function to count the number of 1 in bitboard or to find the first 1.
>>>
>>>Uri
>>
>>I can add that I hesitate what exactly I am allowed to do.
>>I plan to have function that is equivalent o r almost equivalent to Crafty
>>function InputMove(I plan to return 1 and make the move and print the move in
>>notation like g1f3 to some text file in case that the move is legal and to
>>return 0 in case that the move is illegal without making the move).
>>
>>I do not use copy and paste but the code that I wrote for InputMove is clearly
>>influenced by crafty and I cannot say that it is independent.
>
>As a fellow programmer, I can give you my opinion.
>
>I couldn't care less if you copied the complete input move routine from Crafty.
>Copying 10 lines from the evaluation code I would find unacceptable however.
>
>Tony
That is somewhat akin to my feeling. Some functions are finite. They have
distinct inputs and for each distinct input they have a constant output. Move
input is one example. It is just a mapping from ASCII to some internal
representation. Endgame tables are another. Given the same position, _any_
endgame table prober should get the exact same result assuming no bugs.
Any such function could be considered part of a "global toolkit". But that
would not include evaluation code which is obviously a many-to-many mapping
function that depends on what the author wanted to do. Ditto for search and
extensions. Ditto for move ordering.
Whether copying hashing code is OK or not requires some additional thought since
that is almost a 1-to-1 function as well, but there are lots of variations on
replacement strategy, what is stored, etc, which probably means it is not
copyable in general.
We have clearly already reached the point where move input and output are ok to
copy since everyone is using some common set of GUI applications to handle that.
Of course that is not so good IMHO for some GUI applications that handle
opening books, book learning, time allocation, endgame table probes, and so
forth.
Gets muddier and muddier...
>
>>
>>1)I used some names of variables from Crafty's code because I was lazy to try to
>>think about better names.
>>
>>2)I did not know about the function strchr that is used in Crafty before reading
>>crafty's code
>>
>>I give the code that I have so far at the bottom of this post
>>
>>I was interested in the past if there is some free code that I can use for the
>>purpose of reading pgn but it seems that the answer is negative and every
>>programmer needs to reinvent the wheel again even when the tasks are
>>deterministic tasks like reading pgn and getting specific output in format that
>>is more easy for programs and there is no free source file that programmers are
>>allowed to use in order to translate pgn to another format that is more friendly
>>for chess programs.
>>
>>int InputMove(char *text)
>>{
>> int ffile,frank,tfile,trank;
>> int piece,capture,promote;
>> char movetext[64];
>> char* lasttext;
>> if (strlen(text)==0)
>> return 0;
>> if (strlen(text)>=4)
>> if ((text[0]>='a')&&(text[0]<='h')&&
>> (text[1]>='1')&&(text[1]<='8')&&
>> (text[2]>='a')&&(text[2]<='h')&&
>> (text[3]>='1')&&(text[3]<='8'))
>> return makeusermove(text);
>> piece=PAWN;
>> capture=0;
>> promote=0;
>> frank=-1;
>> ffile=-1;
>> trank=-1;
>> tfile=-1;
>> strcpy(movetext,text);
>> /*deleting irrelevant chars from movetext like checks and mates that I ignore
>>and
>> must appear at the end of the text*/
>> if (strchr(movetext,'#'))
>> *strchr(movetext,'#')=0;
>> /*I deleted mate and chars after mate and I continue for checks and
>>promotions*/
>> if (strchr(movetext,'+'))
>> *strchr(movetext,'+')=0;
>> printf(" %d ", strlen(movetext));
>> printf(" %s ",movetext);
>> if (strchr(movetext,'='))
>> {
>> lasttext=strchr(movetext,'=');
>> if (strlen(lasttext)<2)
>> return 0;
>> switch (lasttext[1])
>> {
>> case 'q':promote=QUEEN;
>> break;
>> case 'n':promote=KNIGHT;
>> break;
>> case 'b':promote=BISHOP;
>> break;
>> case 'r':promote=ROOK;
>> break;
>> }
>> *strchr(movetext,'=')=0;
>>
>> }
>> return 0;
>>
>>}
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.