Winboard Forum

http://sourceforge.net/projects/bdi-chess/files/

http://sourceforge.net/projects/protector/

Could somebody compile these for win32? Or are these just non-functionable development in progress versions?

Swaminathan wrote:http://sourceforge.net/projects/bdi-chess/files/

http://sourceforge.net/projects/protector/

Could somebody compile these for win32? Or are these just non-functionable development in progress versions?

Protector is yet another Fruit/TogaII derivate and after reading this I already was too lazy and bored to check the _small
improvements_ in code.

Guenther

According to this thread (http://www.computerchess.info/tdbb/phpBB3/viewtopic.php?f=9&t=394&sid=220ed8b27865543ab01be6114f8d5eb9), Protector is an original engine, and use ideas but not code from Fruit / Toga.

nepossiver wrote:According to this thread (http://www.computerchess.info/tdbb/phpBB3/viewtopic.php?f=9&t=394&sid=220ed8b27865543ab01be6114f8d5eb9), Protector is an original engine, and use ideas but not code from Fruit / Toga.

I suggest you read this thread too :

http://wbec-ridderkerk.forumotion.com/w ... 7-t398.htm

and then make your own opinion

Olivier

On Protector:

Copying ideas is all well and good, but when other people's bugs start showing up in your code, you're doing it wrong.

Swaminathan wrote:http://sourceforge.net/projects/bdi-chess/files/

http://sourceforge.net/projects/protector/

Could somebody compile these for win32? Or are these just non-functionable development in progress versions?

Hi Swami,

You can download Protector builds directly from Sourceforge from here:
http://protector.svn.sourceforge.net/vi ... ector/bin/
Click on the version you are interested in, and then click 'download' for whichever version you want -- you probably want the latest, which is the highest number. These builds can be also be found in the main file download here:
http://sourceforge.net/projects/protector/files/
Inside the tar.gzip file there is a directory called 'bin' where you will find all the builds.
With all the accusations of Protector being a Toga/Fruit clone, I took a (very!) quick glance through some of the code. What I saw looked different to me. But the program design is certainly modeled on Toga/Fruit. Instead of being a direct descendant like a son or daughter, let's call it a cousin. But I'm not sure how distant of a cousin it might be...

I remember reading that bdi-chess doesn't compile, or has severe bugs, or something like that. It's a source code only engine! There's a couple more source-only engines that you can find by looking through the various source code repositories. Most of them are complete messes, probably started by high school kids during the summer, and then quickly abandoned when school restarted. But some of them, like Redqueen Chess Engine, get new commits intermittently and may turn out okay. If you are interested in this stuff you should look at my daily updates page:
http://computer-chess.org/forum/

Ron

Ron Murawski wrote:With all the accusations of Protector being a Toga/Fruit clone, I took a (very!) quick glance through some of the code. What I saw looked different to me. But the program design is certainly modeled on Toga/Fruit. Instead of being a direct descendant like a son or daughter, let's call it a cousin. But I'm not sure how distant of a cousin it might be...

The search is a straight copy of Toga. The differences:

- It doesn't use extensions in full_no_null()
- It doesn't check for 50-move draws in quiescence
- Slightly different extensions and history pruning and an additional condition for nullmove
- A different implementation of futility pruning

Okay, put this way, it sounds completely different. But it's almost identical when you read it. Including at least one bug.

Teemu Pudas wrote:On Protector:

Copying ideas is all well and good, but when other people's bugs start showing up in your code, you're doing it wrong.

From what I read this is not clear at all. Can the same bug not be implemented independently?

Teemu Pudas wrote:

Ron Murawski wrote:With all the accusations of Protector being a Toga/Fruit clone, I took a (very!) quick glance through some of the code. What I saw looked different to me. But the program design is certainly modeled on Toga/Fruit. Instead of being a direct descendant like a son or daughter, let's call it a cousin. But I'm not sure how distant of a cousin it might be...

The search is a straight copy of Toga. The differences:

- It doesn't use extensions in full_no_null()
- It doesn't check for 50-move draws in quiescence
- Slightly different extensions and history pruning and an additional condition for nullmove
- A different implementation of futility pruning

Okay, put this way, it sounds completely different. But it's almost identical when you read it. Including at least one bug.

Well, as he released the source under GPL, his fork is (half) legal - I think if you don't acknowledge the original author you are infringing his copyright, even if you use the same license.

I do like the GPL and GPLed engines, but a clear development team is needed. Too many forks are bad, and Toga suffers from this: Toga lacks a centralised development team, there are endless versions with minimal differences, and its development is probably doomed, as it is unlikely the way it is developed now will bring any improvements, and even if it does, the engine will be ignored by many anyway (in fact, I run quick tests and Protector seems to be much stronger than Toga 1.4.1se, about as strong as Stockfish 1.4, in a win32 OS). Glaurung/Stockfish model is fine with me, and it has worked great so far - Stockfish 1.4 is far stronger than previous versions, and the number of versions is under control so far.

Anyway, I looked at Glaurung and Bison bitboards and it seems they are different from Protector. Any idea if the bitboards are from another program?

edit: added info about 32-bit OS.

Thanks for the detailed response, Ron.

I'm already actively checking your news site. Keep up the good work!

Teemu Pudas wrote:

Ron Murawski wrote:With all the accusations of Protector being a Toga/Fruit clone, I took a (very!) quick glance through some of the code. What I saw looked different to me. But the program design is certainly modeled on Toga/Fruit. Instead of being a direct descendant like a son or daughter, let's call it a cousin. But I'm not sure how distant of a cousin it might be...

The search is a straight copy of Toga. The differences:

- It doesn't use extensions in full_no_null()
- It doesn't check for 50-move draws in quiescence
- Slightly different extensions and history pruning and an additional condition for nullmove
- A different implementation of futility pruning

Okay, put this way, it sounds completely different. But it's almost identical when you read it. Including at least one bug.

Thank you for your interest in Protector, Teemu

It does not make much sense to do a time-consuming check for 50-move-draws in the quiescence search: apart from the first two plies the quiescence search examines exclusively captures and promotions which suspend the 50-move-rule. It's not a bug, it's feature.

Kind regards

Raimund Heid

PS: In fact Toga also doesn't extend in the full_no_null() method (except for check moves), since this method is never called with pvnode=TRUE or with the player to move in check. It sounds completely different. But it's almost identical when you read it

hello Raimund

for your interest Protector take part in our tournament
http://computer-chess.org/doku.php?id=c ... n_1_add_on

best to you