Myracle GUI

[chessica]

Hi Martin,

I’ve noticed that if I run an EPD test twice on the same day, the previous log file gets overwritten.

[mar]

Hi Martin,

I’ve noticed that if I run an EPD test twice on the same day, the previous log file gets overwritten.

hello,
yes that's how it works now
I could mix the current date and time into the filename the way I do with tours without explicit filename

I don't plan a new build soon unless some severe problem is found

[mar]

I know that you generally prefer to avoid manual game evaluations, but in the MessChess tournament mode it’s sometimes unavoidable. Without a manual evaluation, a duel may not finish properly — although this only happens rarely.

But if it doesn’t work, then it doesn’t work. No problem.

yes, the game will eventually end due to a threefold or 50 move rule or TB pos or checkmate or something else,
but I understand that MessChess don't report score so eval-based adjudication won't work in this case, which can be a bit annoying

I'm reluctant to add manual adjudication because that's an external influence (think perhaps star trek and prime directive

now imagine this happens: you click adjudicate draw but before you do the game is adjudicated and the tab closes and switches to another game,

so now you have another problem: you adjudicated the wrong game and it cannot be undone.
now of course I could add a confirmation dialog or something else, but this all feels so hacky and ugly - the game can end naturally while the dialog is up (I cannot pause the active games easily, mostly due to parallelism) - so this opens a giant can of worms

I have a similar opinion on manually changing game results etc. - once the tour is over, people can do whatever they want with the pgn,
once they copy it elsewhere. so far the only (clean) way to influence existing results is replays

[vencenda]

Thank you for the updates, Martin. Today I come with two bugs (I believe) reports and 2 suggestions (as always, feel free to ignore).

Bugs:
1. When you tick "sequential" in book settings, the next time you open the settings it isn't ticked, although the setting itself seems to have been applied.
2. I have set a core budget of 2 (IO core budget of 4). I start a tournament with engines using 1 thread, so 2 games run in parallel. While the tournament is running, on another tab, I click on Analyze, which involves two other 1-thread engines. I would expect that the analysis would not start because of core budget limit. But it starts and the tournament stops after the two active games are finished. But for some time, all 4 engines run together. I'm not sure but I don't think this is the intended behavior.
But even aside from the core budget, those analysis engines were analyzing the games too. Even if they were within budget limit, this doesn't seem normal. They weren't just analyzing the position in the new tab, but also the 2 games being played in the tournament. So each analysis engine was analyzing 3 positions. Does that mean that even more than 4 cores were used?

Suggestions:
1. IMO, it would feel more consistent if "new game" and "new tournament" dialogs had the same options for engines when applicable. Meaning adding threads, hash, syzygy (and maybe now "disable own book" too) options in "new game" dialog too. Seeing that you added hash and threads override in EPD suites, I thought that you may not be against this.
2. I like how you can view previous games of a running tournament with a click, but I'd find it more helpful if eval graphs of those games could be visible too. I wonder if it would be possible for the eval values to be pulled from the PGN when it's loaded, to recreate the eval graph.

[mar]

Thank you for the updates, Martin. Today I come with two bugs (I believe) reports and 2 suggestions (as always, feel free to ignore).

Bugs:
1. When you tick "sequential" in book settings, the next time you open the settings it isn't ticked, although the setting itself seems to have been applied.
2. I have set a core budget of 2 (IO core budget of 4). I start a tournament with engines using 1 thread, so 2 games run in parallel. While the tournament is running, on another tab, I click on Analyze, which involves two other 1-thread engines. I would expect that the analysis would not start because of core budget limit. But it starts and the tournament stops after the two active games are finished. But for some time, all 4 engines run together. I'm not sure but I don't think this is the intended behavior.
But even aside from the core budget, those analysis engines were analyzing the games too. Even if they were within budget limit, this doesn't seem normal. They weren't just analyzing the position in the new tab, but also the 2 games being played in the tournament. So each analysis engine was analyzing 3 positions. Does that mean that even more than 4 cores were used?

Suggestions:
1. IMO, it would feel more consistent if "new game" and "new tournament" dialogs had the same options for engines when applicable. Meaning adding threads, hash, syzygy (and maybe now "disable own book" too) options in "new game" dialog too. Seeing that you added hash and threads override in EPD suites, I thought that you may not be against this.
2. I like how you can view previous games of a running tournament with a click, but I'd find it more helpful if eval graphs of those games could be visible too. I wonder if it would be possible for the eval values to be pulled from the PGN when it's loaded, to recreate the eval graph.

thanks for reporting, both bugs will be fixed in the next build

ad suggestions:
1) no, I'll keep new game it as is, it's not as important as tours

2) this would be nice even for other pgn files with compatible format, but only eval and depth graphs could be extracted this way
it's a bit tricky though because I'd have to parse pgn comments and it would be more of a guesswork than anything else; also decimal point vs decimal comma and similar things to solve, I'll make a note, would be nice to have for sure

[vencenda]

Thanks for considering one of my suggestions, Martin.

Another possible bug:
If you have more than one tabs open and you try to "Paste FEN" or "Paste PGN" in a tab other than the first one, it still loads in the first tab. EDIT: Same with "Setup Board".

[mar]

Another possible bug:
If you have more than one tabs open and you try to "Paste FEN" or "Paste PGN" in a tab other than the first one, it still loads in the first tab.

that's not a bug, that's a feature: paste FEN/paste PGN always pastes into scratchpad, this is by design and intentional

[mar]

build 68 is up:
- fix: sequential checkbox not updated properly in book settings
- fix: analysis start respects scheduler budget
- epd suite log filenames contain actual date and time
- attempt to parse eval, depth and time graphs when loading a pgn
- navigating moves highlights corresponding ply (vertical line) in graph if possible

[Wilhelm]

Hello Martin, thanks for your updates. I have a question about the extra plys in the epd window. I noticed that Myracle set a position to Failed even when the engine has the right move but doesn't reach the number of given extra plys. In a test with set to 50 I got a result of zero. Or if a engine finds the correct move 5sec befor the end of time and I have set 4 extra ply then this is also failed. I think this shouldn't be that way. I get very different results to Fritz and Arena GUI. With high extra ply the engine should use the whole given time to be sure that it doesn't change the move even after 5 or 6 ply. What do you think?

[mar]

Hello Martin, thanks for your updates. I have a question about the extra plys in the epd window. I noticed that Myracle set a position to Failed even when the engine has the right move but doesn't reach the number of given extra plys. In a test with set to 50 I got a result of zero. Or if a engine finds the correct move 5sec befor the end of time and I have set 4 extra ply then this is also failed. I think this shouldn't be that way. I get very different results to Fritz and Arena GUI. With high extra ply the engine should use the whole given time to be sure that it doesn't change the move even after 5 or 6 ply. What do you think?

hi,

that's strange. it's not extra plys but minimum ply needed to accept a solution.
so a solution is marked as found only if the engine reached at least depth <minimum ply> and found one of the best moves
out of curiosity: which engine did you use and which suite and which values you put in minimum ply and max time?