[BUG] Error thrown when move time is negative

[tissatussa]

I successfully compiled your newest v2.0.0 source (after PR 98) on Linux and i played a game in CuteChess GUI. I won the game, but this was because Byte-Knight lost on time, and CuteChess gave the message 'connection stalls' :

It was a 5m+3s game. CuteChess tolerates the human player to exceed the time limit, but not any engine. In games against other engines Byte-Knight uses its time efficiently, upto the very limit i must say, and it never ran out of time. So, i'm surprised in my game it did. I suspect Byte-Knight regards the spent time of the opponent to calculate its own 'remaining' time. If so, you should change that.

Btw. my last move was a blunder and the position is lost.

[replay the game]

[Event "human vs engine"]
[Site "Holland"]
[Date "2025.04.29"]
[Round "?"]
[White "Roelof Berkepeis"]
[Black "Byte-Knight v2.0.0"]
[Result "1-0"]
[ECO "A45"]
[GameDuration "00:12:16"]
[Opening "Queen's pawn game"]
[Termination "stalled connection"]
[TimeControl "300+3"]

1. d4 Nf6 {-0.11/15 17s} 2. Nc3 {21s} d5 {-0.08/17 19s} 3. Bf4 {9.1s}
e6 {-0.02/18 18s} 4. e3 {32s} Bb4 {-0.08/17 18s} 5. Bd3 {39s} c5 {-0.05/18 16s}
6. dxc5 {33s} Nbd7 {0.00/17 14s} 7. Ne2 {47s} Nxc5 {-0.01/17 28s} 8. Bb5+ {3.4s}
Ncd7 {-0.03/18 17s} 9. O-O {5.5s} O-O {0.00/17 15s} 10. a3 {18s}
Be7 {+0.05/17 15s} 11. Bg5 {49s} a6 {+0.27/16 11s} 12. Bxd7 {22s}
Bxd7 {+0.31/15 11s} 13. Ng3 {10s} Rc8 {+0.47/15 8.8s} 14. e4 {24s}
Nxe4 {+1.27/18 9.3s} 15. Ncxe4 {17s} Bxg5 {+1.47/19 17s} 16. Nxg5 {12s}
Qxg5 {+1.48/17 8.5s} 17. f4 {1.2s} Qh4 {+1.42/15 16s} 18. c3 {7.9s}
Bb5 {+1.47/15 7.4s} 19. Re1 {9.5s, Black's connection stalls} 1-0

[ i'm on Xubuntu 22.04 ]

[DeveloperPaul123]

Hello! Thank you for reporting this. Just to be clear, you compiled the engine on this commit 20870c6 correct?

[DeveloperPaul123]

It doesn't seem that byte-knight uses the opponent play time to compute its hard and soft time limits, but I wonder what cutechess is sending the engine if it tolerates a human player exceeding their time.

[DeveloperPaul123]

I found the bug. Fix is in #101. Feel free to give that branch a try if you can. Thanks again for reporting this!

cutechess was sending go wtime -1234 btime 1234 binc 123 winc 123 movestogo 39 when the human player exceeds their time. The UCI parsing library was throwing an error when trying to parse the negative time.

[tissatussa]

Just to be clear, you compiled the engine on this commit 20870c6 correct?

I don't know, i just downloaded the ZIP of the (then) current Main branch, it had the message on top "Merge pull request #98 from .."
Same now : i downloaded the ZIP of the current Main branch, with message "Merge pull request #99 from ..". Using 'Meld' (a GUI for the Linux 'diff' command) i see differences between these 2 source packages, but no sign of changing any code concerning time management !? Can you confirm and/or link to the fixed source package ? I'm not into GitHub commits etc.

[DeveloperPaul123]

Just to be clear, you compiled the engine on this commit 20870c6 correct?

I don't know, i just downloaded the ZIP of the (then) current Main branch, it had the message on top "Merge pull request #98 from .." Same now : i downloaded the ZIP of the current Main branch, with message "Merge pull request #99 from ..". Using 'Meld' (a GUI for the Linux 'diff' command) i see differences between these 2 source packages, but no sign of changing any code concerning time management !? Can you confirm and/or link to the fixed source package ? I'm not into GitHub commits etc.

Ok I see. You will need to download the zip of the fix/time-error branch. The fixes haven't been merged into main yet. https://github.com/DeveloperPaul123/byte-knight/tree/fix/time-error

Or see #101

[DeveloperPaul123]

This has been merged into main and should be fixed.

[tissatussa]

Yes it works : negative human time allowed by Byte-Knight in CuteChess on Linux