Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: General comments

Author: Tord Romstad

Date: 08:37:09 06/10/04

Go up one level in this thread


On June 10, 2004 at 11:00:23, Robert Hyatt wrote:

>On June 10, 2004 at 04:33:21, Tord Romstad wrote:
>
>>On June 09, 2004 at 16:19:10, Robert Hyatt wrote:
>>
>>>Parsing SAN is _not_ hard.
>>
>>Perhaps not extremely hard, but still an unnecessary complication, IMHO.
>>A format like "g1f3" or "Ng1f3" is closer to the internal representation
>>of moves in almost all chess-related programs.  I find it hard to believe
>>that there is a big number of programs which actually stores only the
>>type of piece and the destination square, and adds more information only
>>when it's necessary to avoid ambiguity.
>>
>>By using SAN, you make life more difficult for the programs which produce
>>the files as well as for programs which read the files.  The programs which
>>produce the files are forced to convert it's internal from-to representation
>>of moves to SAN, and the programs which read the files have to reverse the
>>process.  This seems really silly to me.
>
>Notice that I said "I like and prefer SAN myself".
>
>However I proposed that if we do a new standard, we pick _one_ move
>representation and stick with it, and provide conversion utilities to convert it
>to other representations (including san).

Here, I agree 100%.  It is extremely annoying that the xboard protocol,
for instance, does not require some specific move representation.  The
PV in the xboard protocol is essentially just a string, and the GUI has
to *guess* the move format if it wants to do anything more complicated than
just displaying the raw string.

I don't think using SAN is the best solution, but I agree that the most
important thing is to agree about a single representation.

>>SAN is only a good idea when the information is presented to the user.
>>It has no place in file formats which are not designed to be read by
>>humans, nor in engine communication protocols.
>
>
>I don't disagree, except that there are always going to be humans that insist on
>reading the raw data..

Yes, but I don't think making life easier for those who insist on reading
the raw data should be a guideline when designing the standards.

>>>There is public code to do that in the epd kit as well as inside Crafty
>>>itself.
>>
>>I am not sure what the "epd kit" is, but if you are referring to the
>>epd*.c and epd*.h files by Steven J. Edwards which are included in the
>>Crafty source code, they are unusable to the majority of chess programmers
>>because of their gigantic size.  The epd*.* files alone are bigger than
>>the complete source code of my engine, despite the fact that my code has
>>grown very bloated.
>
>
>What does their "size" have to do with being usable?  They don't add megabytes
>to the executable...

Not megabytes, but it wouldn't surprise me if they add 100 KB, which is
already far more than I can accept for something which does not increase
playing strength.

>Ever looked at the _size_ of the shared C libraries you are including without
>knowing???  Hint:  They _are_ megabytes in size.

Yes, but the point is that they are _shared_.  If I were forced to always
deliver statically linked executables, I would have tried to economize my
library usage as much as possible.

Tord



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.