Hi,
if you allow me to ask a question, I wonder if you just copied the Fruit eval (I know you said you changed the terms) ? or did you borrowed other things from Fruit or others programmes, especially concerning the search, extensions, null move,...? This question may look naive but I'm curious to know the relative effect of eval and search in a programme. If you borrowed some other things in the search of open source programmes you should describe it more clearly. At least it is nice to have made the code open source !
My two cents,
Yves
Sloppy 0.1.1 released
Moderators: hgm, Rebel, chrisw
-
- Posts: 750
- Joined: Mon Mar 27, 2006 7:45 pm
- Location: Finland
Re: Sloppy 0.1.1 released
Due to different data structures I couldn't copy the eval code, but I did copy the terms and weights. I also tried some things from Fruit's search: mainly mate-distance pruning, futility pruning, history pruning, history heuristics, and verified null move pruning. Most of these features didn't work well for Sloppy so I dropped them. The search isn't too interesting because just like in every other chess engine, it's just a combination of well-known and tested features.YL84 wrote:Hi,
if you allow me to ask a question, I wonder if you just copied the Fruit eval (I know you said you changed the terms) ? or did you borrowed other things from Fruit or others programmes, especially concerning the search, extensions, null move,...? This question may look naive but I'm curious to know the relative effect of eval and search in a programme. If you borrowed some other things in the search of open source programmes you should describe it more clearly. At least it is nice to have made the code open source !
My two cents,
Yves
util.c also has a couple of functions for bit-twiddling and cross-platform compatibility that weren't written by me. And I use Pradu's magic move generator for getting movemasks for sliding pieces.
I studied a lot of source code, but Fruit was the only thing that was useful to me.
Re: Sloppy 0.1.1 released
Thank you very much for the answer !
When you say that some search features are well known, ok, but they are not at all easy to implement! I agree as you say that some may not work very well... Do you suggest that your eval gives your programme much more strength than the search will give ? I know my question is a complete non sense
You dropped most of the features you tried, but at least you may have search extensions, or a very good SEE?
Yves
When you say that some search features are well known, ok, but they are not at all easy to implement! I agree as you say that some may not work very well... Do you suggest that your eval gives your programme much more strength than the search will give ? I know my question is a complete non sense
You dropped most of the features you tried, but at least you may have search extensions, or a very good SEE?
Yves
-
- Posts: 750
- Joined: Mon Mar 27, 2006 7:45 pm
- Location: Finland
Re: Toga 0.1.1 released
That is the whole point of the GPL:Daniel Mehrmann wrote:Just look on older engines like Quark. I don't wanna know how much hours, days, months or years Thomas spend on his evaluation terms.
1. Developer A spends years developing and optimizing feature X. He wants to contribute to the open source community so he releases X under the GPL.
2. Developer B, who also wants to contribute to open source, needs X in his project, so instead of re-inventing the wheel he uses dev. A's source code for it.
3. Software gets better, both A and B are happy to have achieved their goal.
I copied Fruit's eval terms BECAUSE Fabien spent a lot of time tweaking them, not despite. And I'll be happy if someone does the same for Sloppy.
Obviously the open source community doesn't mix well with the competitive computer chess community.
-
- Posts: 750
- Joined: Mon Mar 27, 2006 7:45 pm
- Location: Finland
Re: Sloppy 0.1.1 released
No problem.YL84 wrote:Thank you very much for the answer !
True. When it comes to search features, I actually found academic papers and discussions here a lot more useful than any source code.When you say that some search features are well known, ok, but they are not at all easy to implement!
There's probably a lot of room for improvement in the search, but I still think that most of Sloppy's future strength improvements are going to be in the evaluation.I agree as you say that some may not work very well... Do you suggest that your eval gives your programme much more strength than the search will give ? I know my question is a complete non sense
Yes, I have check, recapture and promotion threat extensions. I also do late move reductions, but obviously not based on history tables. Sloppy's SEE is nothing special. It does handle en-passant captures and promotions correctly, but not pins.You dropped most of the features you tried, but at least you may have search extensions, or a very good SEE?
-
- Posts: 4556
- Joined: Tue Jul 03, 2007 4:30 am
Re: Toga 0.1.1 released
Unlike commercial programmers that only care about money. I can assure you that if Vas shared his code of Rybka, we'd have a lot of strong engines around (Better software) and probably (Or possibly) an engine stronger than Rybka. These things just cannot happen when the loss of money is unacceptable (Or you live from computer chess).ilari wrote:3. Software gets better, both A and B are happy to have achieved their goal.
Your beliefs create your reality, so be careful what you wish for.
-
- Posts: 6073
- Joined: Sat Apr 01, 2006 9:34 pm
- Location: Scotland
Re: Toga 0.1.1 released
Precise.Ovyron wrote:Unlike commercial programmers that only care about money. I can assure you that if Vas shared his code of Rybka, we'd have a lot of strong engines around (Better software) and probably (Or possibly) an engine stronger than Rybka. These things just cannot happen when the loss of money is unacceptable (Or you live from computer chess).ilari wrote:3. Software gets better, both A and B are happy to have achieved their goal.
He forgot to mention C who gets something better than the rest and is commercial.
I do not mind open source as a concept but I hate and will fight until my dying breath strong open source.
With strong open source we get lots of the same and none of the new.
I want to speak to all those original authors....
We at Arena support you and your efforts. Your efforts are worth more to us than anything copied from whatever source no matter how strong (but that is the norm).
We would say to you not to be discouraged by such plagarism and to continue on your own way. You will never get anything but a friendly and helpful ear from us. We live on originality. That is what makes the computer chess world a better place for both you and those who choose the commercial road.
We want original development.
We do not require what we already have.
It's called progress.
If you feel worried about anything that is going on feel free to tell us as you already have many times. We know that computer chess cannot survive without original works.
Christopher
-
- Posts: 858
- Joined: Wed Mar 08, 2006 9:24 pm
- Location: Germany
- Full name: Daniel Mehrmann
Re: Toga 0.1.1 released
Well, basicly you're right with your GPL views. I'm FSF member and active on some savannah and kde projects.ilari wrote:That is the whole point of the GPL:Daniel Mehrmann wrote:Just look on older engines like Quark. I don't wanna know how much hours, days, months or years Thomas spend on his evaluation terms.
1. Developer A spends years developing and optimizing feature X. He wants to contribute to the open source community so he releases X under the GPL.
2. Developer B, who also wants to contribute to open source, needs X in his project, so instead of re-inventing the wheel he uses dev. A's source code for it.
3. Software gets better, both A and B are happy to have achieved their goal.
I copied Fruit's eval terms BECAUSE Fabien spent a lot of time tweaking them, not despite. And I'll be happy if someone does the same for Sloppy.
Obviously the open source community doesn't mix well with the competitive computer chess community.
But chess is a game, namly a strategy board game as many other games too. It's a competition between programmers and her babies.
So, the basic idea behind GPL doesn't work very well here.
You'll find more closed code programs in contest strategy board games areas and no much GPL engines.
If you ask developers, like Fabien (Fruit/Toga) or Tord (Glaurung), why they are released the versions under the GPL you'll get the answer that they only want that other rookies can learn how to write, or better how it works, a chessprogram and not copy the content in any case.
I can't believe that you didn't know that. However, you should think about it and may you're chance your mind later on.
Best,
Daniel
-
- Posts: 910
- Joined: Wed Mar 08, 2006 9:46 pm
- Location: Plovdiv, Bulgaria
Re: Toga 0.1.1 released
hi Chris
i think you're wrong in your understanding what is original in programming. You said:We do not require what we already have, I'd say you: Nobody gives you by force what you do not require, you're free to not have it. And if you want to say that you already have Sloppy, i have to tell you that this statement is ridiculous.
I think your fighting against evil is not safe for the computer chess community and your acts bring more damage than benefit
There is an old proverb from my village: Along with the dry wood the wet one is burning too. Think about it and about another one (not originating from my village): Road to hell is paved with good intentions.
Beyond Iron Curtain Regards,
Geno
i think you're wrong in your understanding what is original in programming. You said:We do not require what we already have, I'd say you: Nobody gives you by force what you do not require, you're free to not have it. And if you want to say that you already have Sloppy, i have to tell you that this statement is ridiculous.
I think your fighting against evil is not safe for the computer chess community and your acts bring more damage than benefit
There is an old proverb from my village: Along with the dry wood the wet one is burning too. Think about it and about another one (not originating from my village): Road to hell is paved with good intentions.
Beyond Iron Curtain Regards,
Geno
take it easy