When will come a new Toga Engine?

Discussion of anything and everything relating to chess playing software and machines.

Moderators: bob, hgm, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Aleks Peshkov
Posts: 882
Joined: Sun Nov 19, 2006 8:16 pm
Location: Russia

Re: When will come a new Toga Engine?

Post by Aleks Peshkov » Sat Apr 11, 2009 1:19 am

Open/Free Software Foundations made things easier for programmers. Sometimes we have discussions about GPL and other topics, but IMHO all problems regarding open source are solvable.

But chess programmers are also restricted to computer chess tournaments rules. Not all people in the CCC agree that CC programming is sport, but IMHO it seems to be the main stream opinion.

Some emotional persons (mostly non programmers or ex-chess-programmers) blame other participants' in non-sportsmanship and even general non ethical behavior ruling CC game by personal rule set.

As far as I understand there are no clear common CC rule base. We must define and agree strict game rules and tournament conditions or stop blaming persons who actually perform and develop CC sport.

Peter Aloysius
Posts: 57
Joined: Thu Mar 09, 2006 11:53 am
Location: Surabaya, Indonesia

Re: When will come a new Toga Engine?

Post by Peter Aloysius » Sat Apr 11, 2009 11:50 am

Dr.Wael Deeb wrote:
Peter Aloysius wrote:
Dr.Wael Deeb wrote:
Peter Aloysius wrote:
Edward German wrote:Hi all, hello Programmers!

When will we see a brand new Toga Engine? What happens here?

Regards,
Eduard
Someone who create "brand new" toga engine cannot be called programmer, they're ... cloner.
This is totaly incorrect....
A can of worms regards,
Dr.D
_Why incorrect? They know next to nothing except how to use an editor and how to compile.

_They even can't create a strong engine by their own.
_No,you can't change the source and add +100 Elo boost to the original open sourced program if you don't have programming skills....
I for myself can't do that....even if you give me an editor and teach me how to compile....

_Again I disagree....take for example the author of Cyclone....he began to write his own chess engine from scratch and I sure he'll deliver a pretty strong one....

Dr.D
Really? Latest "author" of a glaurung clone only change some values (and off course engine & author name) and the resulted engine is 10 point weaker than original. What a shame... At least he got his name as an author in rating list.

btw, you should try that. Pretty easy actually. Don't forget to put your own name before you compile..

User avatar
Graham Banks
Posts: 33463
Joined: Sun Feb 26, 2006 9:52 am
Location: Auckland, NZ

Re: When will come a new Toga Engine?

Post by Graham Banks » Sat Apr 11, 2009 10:51 pm

As somebody who enjoys testing engines, how am I to know the complete and accurate list of engines that are derivatives?

When I look in certain readme files, I read that this engine or that engine is based on TSCP or some other open source engine, or that the engine has "borrowed" from the open source engine concerned.

Unless there's a pefectly accurate list of derivatives, I'm in a position where I can't really leave some out and not others because that is both unfair and inconsistent.

Or is it just derivatives from stronger open source engines that are under fire here. If so, that would seem unjust and petty.

Regards,
Graham.
My email addresses:
gbanksnz at gmail.com
gbanksnz at yahoo.co.nz

Ryan Benitez
Posts: 719
Joined: Thu Mar 09, 2006 12:21 am
Location: Portland Oregon

Re: When will come a new Toga Engine?

Post by Ryan Benitez » Sat Apr 11, 2009 11:11 pm

While I think it is common sense what engines are based on other engines and how much, your time is yours you should test what ever engines you would like with your own time. Remember though, someone can spend a lot of time and effort to make a new 2200 elo engine. That same person could make a 2850+ elo GPL engine in a day. Both are positive but the person with the 2200 elo engine might feel left out when the community ignores them in favor of a lesser but still positive effort of a different kind.

User avatar
Graham Banks
Posts: 33463
Joined: Sun Feb 26, 2006 9:52 am
Location: Auckland, NZ

Re: When will come a new Toga Engine?

Post by Graham Banks » Sat Apr 11, 2009 11:17 pm

Ryan Benitez wrote:While I think it is common sense what engines are based on other engines and how much, your time is yours you should test what ever engines you would like with your own time. Remember though, someone can spend a lot of time and effort to make a new 2200 elo engine. That same person could make a 2850+ elo GPL engine in a day. Both are positive but the person with the 2200 elo engine might feel left out when the community ignores them in favor of a lesser but still positive effort of a different kind.
Can't argue with most of what you've said, but those who are going to test 2200 elo engines will do so regardless or not.
Some are just interested in the top bracket of engines and will never touch 2200 elo engines whether there are derivatives around or not.

However, I'd still like a complete and accurate list of derivatives of all strengths.
My email addresses:
gbanksnz at gmail.com
gbanksnz at yahoo.co.nz

User avatar
Eelco de Groot
Posts: 4188
Joined: Sun Mar 12, 2006 1:40 am
Location: Groningen

Re: When will come a new Toga Engine?

Post by Eelco de Groot » Sun Apr 12, 2009 4:57 am

Well, what exactly would be the purpose of such a list? For all closed source programs, the derivatives would be clones unless the copyright holder has given permission to make one. I don't know any practical examples of that. Or it would have to be Chess System Tal of which Chris Whittington sold the rights and sources, but I don't know if any CS Tal derivative was ever tested by any of the testing communities. Strelka is a rather unique case of reverse engineering Rybka with added portions of Fruit, Beowulf and probably other programs. CEGT has decided not to test this clone anymore, I don't know of any CCRL testers that still test it for the ratinglist. There are still other versions of Strelka that are probably cases of fraud.

Of the open source programs; this involves mainly Fruit. Crafty clones there are a few but I don't know if there are released versions for which Robert Hyatt I think would have to give permission, there will of course be plenty of private versions, modified Craftys. For Glaurung I know only released versions Stockfish, from Marco Costalba, Smaug by Joona Kiiski, I believe Sloppy was also based on Glaurung but I'm not sure. I don't think it was released with Glaurung's GPL so technically it may not be a derivative of Glaurung. Maybe it stands on its own, I do believe it was published under its own GPL license. Ancalagon was never released, I don't know if it ever will be.

That leaves the Fruit family and of those there are quite a few, and of course I suspect there are many private versions as well. Not everybody wants to be seen with making a private version for testing or experimentation. Hopefully most breaches of the GPL will be caught. But that leaves a list of engines with relevance to the CCRL because tested by the CCRL, for the majority they are first tested by Shaun Brewer through his involvement with Toga testing. For which we are very grateful and I think it is entirely up to him if he wants to do that! Graham you test a lot too mainly for the 40/40 list where Shaun always starts with 40/4. The list is long but nobody I hope is forced to read through it, if he does not want to do that. Most of these versions are private and unless there is a lucky breakthrough will never be released. We do try to limit the number of versions out there, but as an example if somebody is very interested in a 64 bits compile where normally only a 32 bits version would be made, Denis might make an exception for a private special request. No public release will be made in that case.

Only a limited number of new versions made by TDDB developers and others can and will be tested by Shaun.
I believe this list comprises at the moment 129 different versions of Fruit and Fruit derivatives from several authors tested by CCRL in the past through now, starting with original Fruit by Fabien Letouzey of course, and can be watched by clicking here: Fruit Family 40/4 list

I don't know if this is very useful but just to give some data on Graham's question.

Eelco
Debugging is twice as hard as writing the code in the first
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan

Ryan Benitez
Posts: 719
Joined: Thu Mar 09, 2006 12:21 am
Location: Portland Oregon

Re: When will come a new Toga Engine?

Post by Ryan Benitez » Sun Apr 12, 2009 5:39 am

Eelco de Groot wrote:Well, what exactly would be the purpose of such a list? For all closed source programs, the derivatives would be clones unless the copyright holder has given permission to make one. I don't know any practical examples of that. Or it would have to be Chess System Tal of which Chris Whittington sold the rights and sources, but I don't know if any CS Tal derivative was ever tested by any of the testing communities. Strelka is a rather unique case of reverse engineering Rybka with added portions of Fruit, Beowulf and probably other programs. CEGT has decided not to test this clone anymore, I don't know of any CCRL testers that still test it for the ratinglist. There are still other versions of Strelka that are probably cases of fraud.

Of the open source programs; this involves mainly Fruit. Crafty clones there are a few but I don't know if there are released versions for which Robert Hyatt I think would have to give permission, there will of course be plenty of private versions, modified Craftys. For Glaurung I know only released versions Stockfish, from Marco Costalba, Smaug by Joona Kiiski, I believe Sloppy was also based on Glaurung but I'm not sure. I don't think it was released with Glaurung's GPL so technically it may not be a derivative of Glaurung. Maybe it stands on its own, I do believe it was published under its own GPL license. Ancalagon was never released, I don't know if it ever will be.

That leaves the Fruit family and of those there are quite a few, and of course I suspect there are many private versions as well. Not everybody wants to be seen with making a private version for testing or experimentation. Hopefully most breaches of the GPL will be caught. But that leaves a list of engines with relevance to the CCRL because tested by the CCRL, for the majority they are first tested by Shaun Brewer through his involvement with Toga testing. For which we are very grateful and I think it is entirely up to him if he wants to do that! Graham you test a lot too mainly for the 40/40 list where Shaun always starts with 40/4. The list is long but nobody I hope is forced to read through it, if he does not want to do that. Most of these versions are private and unless there is a lucky breakthrough will never be released. We do try to limit the number of versions out there, but as an example if somebody is very interested in a 64 bits compile where normally only a 32 bits version would be made, Denis might make an exception for a private special request. No public release will be made in that case.

Only a limited number of new versions made by TDDB developers and others can and will be tested by Shaun.
I believe this list comprises at the moment 129 different versions of Fruit and Fruit derivatives from several authors tested by CCRL in the past through now, starting with original Fruit by Fabien Letouzey of course, and can be watched by clicking here: Fruit Family 40/4 list

I don't know if this is very useful but just to give some data on Graham's question.

Eelco
I agree with what you said except that Sloppy is an original code engine and a generous GPL contribution by Ilari Pihlajisto. It has influence from Fruit in the tuning of the eval but the code is original. The same guy makes the best chess GUI I have ever seen called Cutechess-cli.

chesstango
Posts: 184
Joined: Thu Mar 09, 2006 9:13 pm

Re: When will come a new Toga Engine?

Post by chesstango » Mon Apr 13, 2009 10:56 pm

For me that's not a big issue that will never end. Some people, scientists, researchers, dont start from scratch. They go on based on public, issued papers of other coleagues and from something already done......
Not big deal in science...... :shock:
Dr. Bernardo Wesler
INFORMATION DOESN'T REPLACE KNOWLEDGE AND KNOWLEDGE DOESN'T REPLACE WISDOM.

Gian-Carlo Pascutto
Posts: 1192
Joined: Sat Dec 13, 2008 6:00 pm
Contact:

Re: When will come a new Toga Engine?

Post by Gian-Carlo Pascutto » Tue Apr 14, 2009 5:29 am

chesstango wrote:For me that's not a big issue that will never end. Some people, scientists, researchers, dont start from scratch. They go on based on public, issued papers of other coleagues and from something already done......
Not big deal in science...... :shock:
True, but you will find that scientists also get quite upset when someone writes a new paper by editing and changing some parts of their old one. Or doesn't properly attribute origins of their research.

mcostalba
Posts: 2684
Joined: Sat Jun 14, 2008 7:17 pm

Re: When will come a new Toga Engine?

Post by mcostalba » Tue Apr 14, 2009 7:18 am

Gian-Carlo Pascutto wrote: True, but you will find that scientists also get quite upset when someone writes a new paper by editing and changing some parts of their old one. Or doesn't properly attribute origins of their research.
For what I have understood this is forbidden with GPL.

Please let me write a practical list of what GPL is for me:


1 - GPL is NOT free !!!!


2- If, due to an exotic mosquito that bites me, I get comletely mad and decide to turn Stockfish in a commercial engine I CAN DO THAT. The GPL license allows me to sell my product, BUT to the people that purchase my product I MUST give the sources along with binaries.


3- If, for the same mosquito as above, I decide to relicense Stockfish as free software I CANNOT do that. I can double license, but Stockfish will be always GPL because is dervied from a GPL software.


4 - If I decide to turn Stockfish in a closed source program I CANNOT do that.


5- GPL chess engines are NOT suitable to compete against closed source ones (commercial or not) because is like to play poker with only your cards on the table: your opponent can see (and if he likes backport) your tricks and you cannot do the same. So in the long term a closed source program will always be stronger then a GPL one. A GPL can develop faster if is able to attract many developers, but once it is released, for its very nature, it cannot keep its position for a long time. I think the main, and absolutely legitimate, reason for people that start from scratch is because they are interested in competition....unfortunatly I am not ;-)


6- If Tord would had licensed Glaurung as a free software, I would had relicensed ANYWAY Stockfish as GPL because I am absolutey not interested in competition. I am much more interested to help in developing a nice engine that can attract new people and new ideas, and can be instrumental in spreading those ideas in the comunity. And GPL fits this nicely while at the same time does NOT fit at all the "my engine will be the strongest out there" wish...at least it could be but just for a few months after the release.



So the bottom line is that in my opinion GPL is rather a restrictive kind of license especially in this world where competition and past tradition both yield naturally to very different kinds of software license.

At the same time I think that GPL will have a very deep, although long term, influence in chess engines, and especially in the way of thinking of people around this world. And I strongly believe this new way of thinking is a good thing and something that makes sense to work for.

Post Reply