TalkChess.com

Well, I have never seen Tim's code for this. By the time I got involved in WinBoard it had already been replaced by code from Tord.

Fact is that it is detrimental to have incompatible systems co-exist, and that it would be logical to prefer the fundamentally correct standard over the broken one, even if it is only 'slightly broken'.

That reminds me of math class, where the teacher always used to say: "Nearly right! Thus WRONG!!!".

Interesting. It turns out WinBoard recognizes two undocumented 'feature' commands from the engine: oocastle and fen960. According to the comment Tord is responsible for adding those. oocastle is by default true, and causes castlings to be sent as O-O or O-O-O in FRC. I suppose WB engines that would send oocastle=0 would get castlings in KxR format. fen960 is false by default, and currently does not affect anything.

None of this of course will affect how UCI engines are handled.

It comes in very handy that WinBoard can be set to send KxR castlings: it means I can relieve UCI2WB of the task to convert those. (Which would be pretty hard, as UCI is game-state unaware, and would thus have no idea where the King and Rook are located.) I am pretty sure that on input XBoard will understand them too. So all I have to do is let UCI2WB send feature oocastle=0 when the engine does have a UCI_Chess960 option, and leave it at the default 1 when the engine hasn't (assuming that it uses Arena protocol). In the latter case the castling field should be forced to KQkq.

hgm wrote:Fact is that it is detrimental to have incompatible systems co-exist, and that it would be logical to prefer the fundamentally correct standard over the broken one, even if it is only 'slightly broken'.

You are completely right.

Would also be nice to have a Chessbase GUI that supports Winboard, or has at least a correct UCI implementation. But years ago I gave up the hope for that.

hgm wrote:
Fact is that it is detrimental to have incompatible systems co-exist, ".

Well, you are supporting both Winboard and UCI are you not ?

Fact is it is a common occurrence

- Windows and Linux
- UCI and Winboard
- Namimov, Gaviota and Sysgy tablebases
etc

Android and iOS and Windows

A fact of life, it is called freedom of choice. How boring if "incompatible" cars didn't exist and we all drove the same thing.

Modern Times wrote:Well, you are supporting both Winboard and UCI are you not ?

Sure. But that doesn't mean the existence of UCI isn't detrimental. The very fact that I had to support it is living proof for how detrimental its existence is! And why? Basically just to degrade the performance by sending more data to the engine, and to prevent that engines could resign and make or accept draw offers...

A fact of life, it is called freedom of choice. How boring if "incompatible" cars didn't exist and we all drove the same thing.

You are confusing 'compatible' with 'identical' here. In fact all cars are compatible: They have the brake and accelerator pedals in the same relative position, use similar steering wheels, have the gear stick next to your seat, etc. What you are advocating is that there should be more cars that have the gas handle hanging from the roof so that you would have to pull it down with your right hand to accelerate, where you would steer with your feet like in an airplane, and where you would have buttons in front of you to select the gear. And of course for which you would need a different license to drive in...

Do you really think that would be a good idea? Because that is what incompatibility means. That the same drivers license could not be used in every car...

See http://www.hans-egebo.dk/Jokes/cars_vs_computers.htm for more along your line of thought.

Modern Times wrote:

hgm wrote:
Fact is that it is detrimental to have incompatible systems co-exist, ".

Well, you are supporting both Winboard and UCI are you not ?

Fact is it is a common occurrence

- Windows and Linux
- UCI and Winboard
- Namimov, Gaviota and Sysgy tablebases
etc

Android and iOS and Windows

A fact of life, it is called freedom of choice. How boring if "incompatible" cars didn't exist and we all drove the same thing.

Hello Ray,

You are perhaps unaware that your argument nullifies your major thesis?

I can start up Linux and Windows on the same physical architecture. In fact, Linux will run like lightning on systems that Windows would throttle to death. I can even run them side by side on the same machine.

Engines can handle multiple tablebase formats. So can guis. They are not incompatible at all.

UCI and Winboard are alternative protocols, not life or death competitors in a deathmatch arena. In fact, it was not unusual "back in the day" to have engines that supported both protocols in the same executable!

There is no incompatibility.

Now, a poisonous atmosphere of methane or chlorine is incompatible with an iron based metabolism depending on oxygen. Likewise, unassisted diving is incompatible with exploring the bottom of the Marianas Trench.

Freedom of choice implies compatibility. Or the free choice persons will find themselves perpetually frustrated as they encounter guis doing their own thing with the same protocol etc.

You are free to worship the being of your choice, but if that worship involves the slaughter of pet poodles and cats, then you are going to have problems in a number of neighbourhoods. That is an example of incompatibility and freedom of choice.

Your examples do not, to me, make that point.

Later.

On a more productive note: what would be an engine that plays according to UCI Arena960 protocol, that I can test on? Preferably one that exists as Linux version. And where could I download it?

I have patched UCI2WB now so that it will report it supports variants fischerandom and normal if the engine reports a UCI_Chess960 option, or if WinBoard was already set to FRC when the engine was started ('unannounced FRC'). In the former case it sets WinBoard to use KxR notation for castling by sending feature oocastle=0. In the latter case it will replace the castling field of any incoming FEN by KQkq. In theory that should do it. But I would need something to test it on.

If it works I can change the USI/UCCI checkbox in WinBoard to USI/UCCI/Arena960.

hgm wrote:On a more productive note: what would be an engine that plays according to UCI Arena960 protocol, that I can test on? Preferably one that exists as Linux version. And where could I download it?

...

I think ... Hermann.

hgm wrote:On a more productive note: what would be an engine that plays according to UCI Arena960 protocol, that I can test on? Preferably one that exists as Linux version. And where could I download it?

Gothmog, but it's only for windows.

http://www.rwbc-chess.de/download.htm

I think Chispa 4.0.3 ?