Problems with running the engine

[SzotsGabor]

For me it does not move at all under Arena. Under Cute Chess it does move but exceeds time soon. Both with traditional and Fischer clock.

Here is a debug from Arena:

2021-09-11 10:38:33,932-->1:quit 2021-09-11 10:38:34,3231-------------------Starting engine 1 Walleye 1.1.0 64-bit------------------- 2021-09-11 10:38:34,3231Configured Engine 1 Type: UCI 2021-09-11 10:38:34,3231Engine 1 dir: E:\Sakk\UCI-engines\Walleye 2021-09-11 10:38:34,3231Engine 1 commandline: E:\Sakk\UCI-engines\Walleye\Walleye_1.1.0-x64.exe 2021-09-11 10:38:34,4321Child Process Prio Adj: PID 1216 conhost.exe 2021-09-11 10:38:34,4321Engine 1 ProcessID: 7064 2021-09-11 10:38:34,4321Engine 1 Prio:32 ThreadPrio:0 2021-09-11 10:38:34,463-->1:uci 2021-09-11 10:38:34,463<--1:id name Walleye 2021-09-11 10:38:34,463<--1:id author Mitchel Paulin 2021-09-11 10:38:34,463<--1:uciok 2021-09-11 10:38:34,4631Child Process Prio Adj: PID 1216 conhost.exe 2021-09-11 10:38:34,463-->1:isready 2021-09-11 10:38:34,526<--1:readyok 2021-09-11 10:38:38,729----------New game---2021-09-11 10:38:38,729 Szo ------------- 2021-09-11 10:38:38,729Loading book: E:\Sakk\Arena2\Books\IM_4mvs.abk 2021-09-11 10:38:44,9621Start calc, move no: 1 2021-09-11 10:38:44,9621Main Book Move IM_4mvs.abk: c7c5 2021-09-11 10:38:45,9471Start calc, move no: 3 2021-09-11 10:38:45,947-->1:ucinewgame 2021-09-11 10:38:45,947-->1:isready 2021-09-11 10:38:50,056-->1:position startpos moves e2e4 c7c5 e4e5 2021-09-11 10:38:50,056-->1:go wtime 121484 btime 121000 winc 1000 binc 1000

At this point I checked with Process Explorer and the engine was not even in memory.

[MitchelPaulin]

The engine should also produce a log.txt file, usually in the same directory it was run from. Could you also send the file?

[SzotsGabor]

ENGINE << uci ENGINE >> id name Walleye ENGINE >> id author Mitchel Paulin ENGINE >> uciok ENGINE << isready ENGINE >> readyok ENGINE << ucinewgame ENGINE << isready

[MitchelPaulin]

Thanks for the log file, it helped track down the problem to the fact that after the engine received a ucinewgame command it expected to then receive a position command immediately following.

After reading the UCI protocol again it looks like it is possible that it could receive an isready command before this. The most recent commit should now handle this case, could you let me know if rebuilding the engine from master fixed this for you?

[SzotsGabor]

Yes, it moves now. But I don't know if Fischer clock is implemented, I played one game, I was totally lost but it exceeded time. TC was 2 minutes + 1 seconds.

[MitchelPaulin]

Yes, the engine does not know anything about the current time control. It will always search to a depth of 15 (5 normal and a max of 10 quiescence) regardless of whether it has 5 seconds or 5 hours left on its clock. This depth was chosen since they seemed to work well in a 5 minute or so time control playing against my friends, at least on my PC.

Its on the list of things to do along with adding a transposition table and a make and unmake routine to speed up move generation, but as this is a weekend project not sure when I'll be able to get around to it.

Regardless thank you for reporting the bug :). I'll make a 1.1.1 release soon

[MitchelPaulin]

https://github.com/MitchelPaulin/Walleye/releases/tag/1.1.1

[SzotsGabor]

https://github.com/MitchelPaulin/Walleye/releases/tag/1.1.1

Yes, the engine does not know anything about the current time control. It will always search to a depth of 15 (5 normal and a max of 10 quiescence) regardless of whether it has 5 seconds or 5 hours left on its clock. This depth was chosen since they seemed to work well in a 5 minute or so time control playing against my friends, at least on my PC.

Its on the list of things to do along with adding a transposition table and a make and unmake routine to speed up move generation, but as this is a weekend project not sure when I'll be able to get around to it.

Regardless thank you for reporting the bug :). I'll make a 1.1.1 release soon

Yes, the engine does not know anything about the current time control. It will always search to a depth of 15 (5 normal and a max of 10 quiescence) regardless of whether it has 5 seconds or 5 hours left on its clock. This depth was chosen since they seemed to work well in a 5 minute or so time control playing against my friends, at least on my PC.

Its on the list of things to do along with adding a transposition table and a make and unmake routine to speed up move generation, but as this is a weekend project not sure when I'll be able to get around to it.

Regardless thank you for reporting the bug :). I'll make a 1.1.1 release soon

Thanks for the quick fix. I am looking forward to further developments.

For the moment I must skip testing this version for our rating list (CCRL). All testers and testing groups I know require the engine to be able to use basic time control schemes so if you would like to see your engine appear in one of them, that is a must.

I am watching your github page from now on and as soon as there will be a breakthrough I will announce it at the Computer Chess Club.