Buggy engines and building a bug base

[Matthias Gemuh]

I guess I will drop SAN support. The console mode will still accept it, but I am starting to see that there is no point in using it with the winboard/xboard protocol, particularly when there are GUIs that do not follow it.

Miguel

Good decision, Miguel
BTW, what does "Gaviota" mean ?

Matthias.

[Matthias Gemuh]

The all-knowing programming god Donald Knuth said that "premature optimization is the root of all evil". Have you actually profiled ChessGUI with and without the check and mate notation? I'm curious because I've done a lot of profiling with Cute Chess (a work-in-progress GUI), and full SAN support is not even close to being a bottleneck. Updating the graphics on the visual chessboard takes way more cpu cycles, and even that's not significant if it's done right.

My compiler (Borland C++ Builder 5) has no profiler.

Matthias.

[Matthias Gemuh]

WinBoard/Arena/ChessGUI/WB2UCI do a lot of weird things to support as many "buggy" engines as possible.

Matthias.

In my opinion this is one of biggest problems with Winboard engines and GUIs. If an engine fails to obey the chess protocol, the GUI should simply reject the engine. This would keep the GUI codebase shorter, simpler and less buggy, and it would help engine authors as well, because it would be easier to detect bugs and non-standard behavior.

I agree. It is old WinBoard that started the mistake of supporting protocol-violating engines (at least 10% of all).
If your GUI is strict, people will attack you with:
"WB Engines A,B,C all run fine under WinBoard, but not under your GUI. You must be doing something wrong.".

Matthias.

[Ilari Pihlajisto]

I agree. It is old WinBoard that started the mistake of supporting protocol-violating engines (at least 10% of all).
If your GUI is strict, people will attack you with:
"WB Engines A,B,C all run fine under WinBoard, but not under your GUI. You must be doing something wrong.".

Matthias.

For Cute Chess we've decided not to support anything that violates the protocol. Instead we tried to make our GUI detect problems as early and as accurately as possible. Currently the only engine that doesn't work (that I've tried) is Phalanx, which fails even at the most basic task of sending the moves in the right format. Another problem that I thought about, but decided not to fix, was that Crafty's resignation command wasn't accepted due to an unprintable character at the beginning of the command. These kinds of bugs should be fixed in the engine, not in the GUI.

[Matthias Gemuh]

For Cute Chess we've decided not to support anything that violates the protocol. Instead we tried to make our GUI detect problems as early and as accurately as possible. Currently the only engine that doesn't work (that I've tried) is Phalanx, which fails even at the most basic task of sending the moves in the right format. Another problem that I thought about, but decided not to fix, was that Crafty's resignation command wasn't accepted due to an unprintable character at the beginning of the command. These kinds of bugs should be fixed in the engine, not in the GUI.

A few common promotions to Queen are: "h7h8/Q", "h7h8", "h7h8,q"
A few castlings are: o-o-o, 0-0-0 (instead of O-O-O)
Even strong Scorpio sends moves as: "move Ng1-f3" !
... etc.

Your GUI may end up excluding a lot of engines.
I will wait and see if the mob shall crucify you.

Matthias.

[Ilari Pihlajisto]

A few common promotions to Queen are: "h7h8/Q", "h7h8", "h7h8,q"
A few castlings are: o-o-o, 0-0-0 (instead of O-O-O)
Even strong Scorpio sends moves as: "move Ng1-f3" !
... etc.

Your GUI may end up excluding a lot of engines.
I will wait and see if the mob shall crucify you.

Matthias.

Then I'll crucify them back.
Fortunately Scorpio is open source, so I might just send a patch to Daniel. I think we should put more pressure on engine developers to obey the protocol.

[Daniel Shawul]

I wanted a cooler mover notation than g1f3 but did not want to implement SAN. The sub-standard SAN worked for me
and I didn't know it is a violation of the protocol. Does sending "feature san = 0" in the beginning mean that you stick to
the standard notation for the output as well ? Anyway I will switch to the standard e2e4 notation from the next version onwards.
Thanks for cutechess-cli. Will it have ICC support? I want to run some games in the background without
my boss noticing

Daniel

[Gábor Szots]

BTW, what does "Gaviota" mean ?

Sea gull.

[Ilari Pihlajisto]

I wanted a cooler mover notation than g1f3 but did not want to implement SAN. The sub-standard SAN worked for me
and I didn't know it is a violation of the protocol. Does sending "feature san = 0" in the beginning mean that you stick to
the standard notation for the output as well ?

The specs only say that with "san=0" Xboard will send moves to the engine in long algebraic notation. But it's still probably best to use the same format for input and output.

Anyway I will switch to the standard e2e4 notation from the next version onwards.

Thanks, that's good to hear.

Thanks for cutechess-cli. Will it have ICC support? I want to run some games in the background without
my boss noticing

When Cute Chess gets ICC support, then I guess cutechess-cli will get it too. But that's going to take a while. We still have the following tasks to deal with first:
- Implement a human player
- Add graphical widgets for engine output, move list, UCI configuration, etc.
- Add menus and dialogs for stuff that's already implemented in our chesslib, like opening books, time controls, variants, etc.
- Tournament support (Gauntlet, Round Robin, etc.)
- Fix the QProcess problems on Windows
- Better error handling, unit tests, documentation, etc.

[Dave Gomboc]

The all-knowing programming god Donald Knuth said that "premature optimization is the root of all evil". Have you actually profiled ChessGUI with and without the check and mate notation? I'm curious because I've done a lot of profiling with Cute Chess (a work-in-progress GUI), and full SAN support is not even close to being a bottleneck. Updating the graphics on the visual chessboard takes way more cpu cycles, and even that's not significant if it's done right.

My compiler (Borland C++ Builder 5) has no profiler.

Matthias.

So, you're basically admitting that the actual reason you don't generate proper SAN isn't actually due to a performance issue -- it's because you just don't feel like it.

Miguel: You should turn your SAN support back on. It will apparently work with CuteChess.

Dave

[Ilari Pihlajisto]

So, you're basically admitting that the actual reason you don't generate proper SAN isn't actually due to a performance issue -- it's because you just don't feel like it.

Miguel: You should turn your SAN support back on. It will apparently work with CuteChess.

Dave

Just to satisfy my curiosity I measured Cute Chess' SAN generation performance: On my PC it takes an average of 0.0022585 milliseconds to create one SAN move. So to waste even one second of precious CPU time I have to generate almost half a million SAN moves. Doesn't seem like a problem. And this is with code that was optimized for readability, not for speed.

[Miguel A. Ballicora]

The all-knowing programming god Donald Knuth said that "premature optimization is the root of all evil". Have you actually profiled ChessGUI with and without the check and mate notation? I'm curious because I've done a lot of profiling with Cute Chess (a work-in-progress GUI), and full SAN support is not even close to being a bottleneck. Updating the graphics on the visual chessboard takes way more cpu cycles, and even that's not significant if it's done right.

My compiler (Borland C++ Builder 5) has no profiler.

Matthias.

So, you're basically admitting that the actual reason you don't generate proper SAN isn't actually due to a performance issue -- it's because you just don't feel like it.

Miguel: You should turn your SAN support back on. It will apparently work with CuteChess.

Dave

It is back on
Seriously, I added a switch to turn it on or off in the ini file. I may probably make the default "off".

Miguel
PS: Dave, I answered your question about tuning in CCC. Did you see it?