My gift to the community – a much better cutechessGUI

[RubiChess]

I would suggest to get a Github account, create an issue in the Yoshie2000 fork and provide some useful information there like the tournament file.

[Gabor Szots]

I would suggest to get a Github account, create an issue in the Yoshie2000 fork and provide some useful information there like the tournament file.

I already have a github account. Just thought this place is better. Many times I tried to reach authors via their github pages and for some of my issues I am still waiting for an answer.

At the moment my problem is to detect some pattern. I resumed my tournament this morning and it is running fine so I dare not stop it lest I can't resume it which I cannot afford because it is for the CCRL blitz list.

[Gabor Szots]

I would suggest to get a Github account, create an issue in the Yoshie2000 fork and provide some useful information there like the tournament file.

Just had a look, there is no 'Issues' there. And, as I can see, your Jan 25 pull request has been unanswered yet.

[pohl4711]

I think the new GUI is really nice and appreciate the effort to provide it to the community. One minor nit/question does not relate to the GUI but to the cutechess-cli binary. If you decide to use it directly (instead of the GUI), you get two warnings which I assume are related to the new abilities of the GUI to save tournaments (and perhaps another feature?). The warnings are:

Code: Select all

Warning: QFSFileEngine::open: No file name specified
Warning: Tournament::saveTournament(): Could not open file for writing:  ""

I wonder if it's possible to update the cutechess-cli --help output to document what is needed to provide the new parameters it's looking for.

If it's not, no worries. The GUI is very nice!

I use the new cutechess-cli for my UHO-Top15 Ratinglist testruns - no problems here at all. Just copy all the .dll files and other helpfiles from the cutechessGUI folder into the folder, where you use cutechess-cli.

[RubiChess]

I would suggest to get a Github account, create an issue in the Yoshie2000 fork and provide some useful information there like the tournament file.

Just had a look, there is no 'Issues' there. And, as I can see, your Jan 25 pull request has been unanswered yet.

I talked to Patrick via Discord about that. He doesn't want to get into a maintainer role for Cutechess so he refuses to activate the issue feature on his fork. But he will look into gauntlet problems which were unknown because Stefan doesn't use gauntlets.
And he apologized for missing my PR

Regards, Andreas

[Gabor Szots]

I would suggest to get a Github account, create an issue in the Yoshie2000 fork and provide some useful information there like the tournament file.

Just had a look, there is no 'Issues' there. And, as I can see, your Jan 25 pull request has been unanswered yet.

I talked to Patrick via Discord about that. He doesn't want to get into a maintainer role for Cutechess so he refuses to activate the issue feature on his fork. But he will look into gauntlet problems which were unknown because Stefan doesn't use gauntlets.
And he apologized for missing my PR

Regards, Andreas

Thanks, Andreas.

Some news on my part: I am just running a gauntlet using 2 seeded engines and concurrency set to 2, rounds=32, encounter per match=1. With theses premises the gauntlet is running flawlessly. I have just checked whether the same seeded engine plays itself occassionally but I did not find such an occurrence.

[Modern Times]

I talked to Patrick via Discord about that. He doesn't want to get into a maintainer role for Cutechess so he refuses to activate the issue feature on his fork. But he will look into gauntlet problems which were unknown because Stefan doesn't use gauntlets.
And he apologized for missing my PR

Regards, Andreas

It's sad, and once again it ends like this - tragedy.

From Stefan's original post, I thought this was only ever meant to be a one-off set of improvements, not an ongoing project.

[Overwrite3218]

I had a question about tablebases: On the sp-cc.de page the author of "UHO-Top15 Engines Ratinglist" talks about his test conditions for Cutechess and says "Tablebases: None for engines, 5 Syzygy for cutechess-cli". I'm trying to understand this and why reporting tablebase hits reporting in Cutechess seems inconsistent from engine to engine. It seems to work for all Stockfish engines from 17.1 on and Obsidian 16.0, but not Plentychess 7.0.37, Reckless 0.9.0, or Alexandria 9.0. I think one of those that I'm not getting tbhits from doesn't have a Syzygy path. Also, is the reason the sp-cc.de page the author doesn't use Syzygy for the individual engines because not all of them use tablebases?

NOTE: I'm pretty new to Cutechess having run many tournaments in Aquarium. I can't get Plentychess to work in Aquarium's tournament mode and setting up custom opening libraries is possible, but a real pain to setup. Cutechess was a breeze for both!

[peter]

If I yet may return to the point of the font size. With my TV- screen I still can't choose engines for players in New Tournament without changing Windows- resolution, so I once in while asked Claude about it, here's what the bot told me after my question

why the "New Tournament" dialog doesn't fit on my screen (a TV used as monitor, unusual resolution) even with "Global font size" set to the minimum (8).

The dialog's top-level layout in projects/gui/ui/newtournamentdlg.ui has:

Code: Select all

xml
<layout class="QVBoxLayout" name="verticalLayout">
 <property name="sizeConstraint">
  <enum>QLayout::SetFixedSize</enum>
 </property>

QLayout::SetFixedSize locks the dialog to exactly its computed sizeHint() and makes it non-resizable — no dragging the edges, no maximizing.
...
Lowering the global font size shrinks the computed sizeHint somewhat, which is presumably why it fixed things for others, but it doesn't help once the fixed size is still larger than the available screen — there's no fallback to make the window scrollable or shrinkable below that computed minimum.

Removing the sizeConstraint/SetFixedSize block (3 lines) makes the dialog behave like a normal resizable QDialog, so users could just drag it to fit their screen.

Claude offered to make a patch for me too, but I would have needed extensive help for compiling the changed sources too, so I thought I'd just tell about that here, maybe yet other users still have the same problem too, and just in case a new one update was planned sooner or later anyhow