Polyglot1.4.30b

[H.G.Muller]

Now it's Winboard's turn again for further improvements in supporting UCI engines ...

Well, tell us what you need... (Yes, I know, a scroll-bar in the Engine-Settings dialog window... )

[F.Huber]

Well, tell us what you need... (Yes, I know, a scroll-bar in the Engine-Settings dialog window... )

Oh god, for writing my complete wishlist it would certainly be better to start a new thread ...

But ok, some of the most urgent wishes:
1) wider combo boxes (how should you chose an option when you can't even read its full name)
2) making these combo boxes work under Win98/2k (i.e. to really see the list)
3) true support of spin boxes
4) as you already mentioned above: better arrangement of the settings window (either smaller or a scroll bar)
5) a possibility to change/save the defaults at least for the Polyglot options
(I wanted to make 'RepeatPV = false' the default, but entering this line in 'polyglot_1st.ini' didn't work at all, because WB creates a new default one at every start)
6) .. 100) well, I keep them for later ...

BTW, one VERY comfortable feature (which has absolutely nothing to do with UCI support) would be an additional "Start/Stop Engine" button (or at least a function key) in WB, which combines several other functions:
a) when it's not the engine's turn, it should execute either "Machine White" or "Machine Black" (depending on which side is to move next)
b) when the engine is thinking ist should stop it (i.e. "Move now")
c) and maybe a right-click on this button could work as "Abort".
So you would have 4 different function with a single button-click (most GUIs have such a Start/Stop button)

[H.G.Muller]

Well, (2) is obviously at the top of my list; any feature we add should work under all Windows versions we officially support. But no progress there, yet. When I try to open the combos on my Win2k I do see the black line at the bottom thicken. But my hypotheis that other controls were covering it seems invalid: even when the line below it is empty there is no list. It is like it refuses to draw beyond the orignal size limits I defined for it (which in Win XP only define the size of theunselected box.

As to (5): My original design was that Polyglot would be responsible for saving the ini file (just like the engine itself would be responsible for saving its ini file if it was a WinBoard engine); to this end it it could define WB options "Polyglot Save" and "Polyglot IniFile". You cold then either start through the -fUCI option, using the WB-created polyglot_1st.ini, and change the name in the menu to another one before saving, or you could start up the "old" way, using "polyglot engine.ini" as engine command with a hand-made (or externally provided) engine.ini. Then you could press the Save button in the Engine Settings dialog to update that ini file (without WB later overwriting it), without having to bother changing the IniFile name.

This design was based on the evolutionary approach: engine settings are stored in polyglot.ini files in the old method, so let's leave them there for maximum backward compatibility, only adding the possibility now to edit these files from the GUI menu.

[F.Huber]

Well, (2) is obviously at the top of my list; any feature we add should work under all Windows versions we officially support. But no progress there, yet. When I try to open the combos on my Win2k I do see the black line at the bottom thicken. But my hypotheis that other controls were covering it seems invalid: even when the line below it is empty there is no list. It is like it refuses to draw beyond the orignal size limits I defined for it (which in Win XP only define the size of theunselected box.

Could you post here how exactly you're calling this WinAPI function to display these combo boxes?
So I can compare your usage with my Win32API documentation and maybe find any differences for older Windows versions.

[F.Huber]

I think I've found the reason for your combobox problem:

With the help of a small hackertool ('Customizer') I could see that you're using 21 for the height of your comboboxes!
Of course that's much too small, because this value is not the height of the closed control but of the opened box (i.e. with the dropdown-list).
With 'Customizer' it's also possible to temporarily modify those values, and after I set it e.g. to 100 the list drops down without any problems under Win98.

Maybe WinXP doesn't care so much about this value and automatically choses a higher default value, when this height is defined too small, and so this problem doesn't happen here!?

Of course I can't guarantee if it really works (at least it did with my small tool here!), but since it's not much work for you to use higher values you should give it a try ...

[Edit] BTW, make this value large enough (even 500 won't hurt), because this is the maximum height, i.e. if the list is shorter then only the needed height is used.

[H.G.Muller]

Currently I do it like this

Code: Select all

typedef struct {
    DLGITEMTEMPLATE item;
    WORD code;
    WORD controlType;
    wchar_t d1, data;
    WORD creationData;
} Item;

struct {
    DLGTEMPLATE header;
    WORD menu;
    WORD winClass;
    wchar_t title[20];
    WORD pointSize;
    wchar_t fontName[14];
    Item control[MAX_OPTIONS];
} template = {
    { DS_MODALFRAME | WS_POPUP | WS_CAPTION | WS_SYSMENU | DS_SETFONT, 0, 0, 0, 0, 295, 300 },
    0x0000, 0x0000, L"Engine #1 Settings ", 8, L"MS Sans Serif"
};

void AddControl(int x, int y, int w, int h, int type, int style, int n)
{
    int i;

    i = template.header.cdit++;
    template.control[i].item.style = style;
    template.control[i].item.dwExtendedStyle = 0;
    template.control[i].item.x = x;
    template.control[i].item.y = y;
    template.control[i].item.cx = w;
    template.control[i].item.cy = h;
    template.control[i].item.id = 2000 + n;
    template.control[i].code = 0xFFFF;
    template.control[i].controlType = type;
    template.control[i].d1 = ' ';
    template.control[i].data = 0;
    template.control[i].creationData = 0;
}


Code: Select all

   case ComboBox:
       AddControl(x, y+1, 95, 9, 0x0082, SS_ENDELLIPSIS | WS_VISIBLE | WS_CHILD, i);
       AddControl(x+95, y-1, 50, 11, 0x0085, CBS_AUTOHSCROLL | CBS_DROPDOWN | WS_VISIBLE | WS_CHILD | WS_TABSTOP, i+1);
       break;


That means I even specify a height of 11 for the combobox. But we already concluded that this was not pixels.

I guess you are right, and that I really should have seen that myself by looking at the existing code. The fixed dialog templates in the winboard.rc resource file have as definition for the comboboxes in the startup dialog:

Code: Select all

    COMBOBOX        OPT_ChessEngineName,17,74,194,129,CBS_DROPDOWN |
                    CBS_AUTOHSCROLL | WS_VSCROLL | WS_TABSTOP


This specifies a height of 129, which is much larger than they look when not opened, and I should have noticed that. What I don't get then is how it decides the height of the combobox when it is not opened. Apparentl the user does not have to specify that at all???

I will definitely try to put a much larger number there.

[F.Huber]

What I don't get then is how it decides the height of the combobox when it is not opened.

From what I've read anywhere the height of the closed box is determined from the fontsize of its text entries.

[H.G.Muller]

OK, new version again here.

I set the height to 500. It works as before under XP. I still have to transfer it to my Win2k PC to test it there.

[edit] Yep, works!

[F.Huber]

[edit] Yep, works!

And in Win98 too.

One other short (completely different) question:
Even if I switch OFF "Info in Move History" in the WB options, when I'm going through the moves of a just played game with the navigation buttons, then on EVERY engine move there pops up an small (comment) window with then engine's evaluation, which has to be closed manually - very annoying.
Is this a bug or is it intended?

And one other thing (don't know if it's the fault of Winboard or Polyglot?), that seems to happen only with my own engine ChestUCI (but only in WB+PG):
The loading of ChestUCI takes quite a long time (several seconds), and this also on my faster desktop!?
I have no idea what could cause this, it doesn't happen in other GUIs (e.g. Arena) and also not with other UCI engines.
Any idea?
Or is there any way to get a log already from this loading process to see what happens?

[H.G.Muller]

The comment window can be suppressed by the option /autoCommentPopup=false. It is saved in the winboard.ini file, but unfortunately the compiled-in default for this option has always been "true". There is no menu item yet to control this feature from the menus, you would have to give it on the command line (or type it behind your engine name or in the field "additional options" of the startup dialog, or edit it in the winboard.ini. Good thing is that you only hve to do that onece. In the installer pckge that we are going to distribute I will probably set this option to "false" in the ini file we supply, as usually it is indeed very annoying.

I am not sure about the startup time; WinBoard has a timeout period for waiting on the engine sending feature commands, but this only noticeable in engines that do not send feature done=1, and I think Polyglot does that. It could be that Polyglot employs some timed-out waiting towards the engine bfore sending done=1 to WB, waiting for something it does not get in this case. Michel would have to answer that.

To have WinBoard make a log you should start with the option /debug (e.g. in the "additional options" field). Polyglot can make a log, but I don't think logging is on by default, and by the time you can switch it on you are likely too late to see the startup. So you would have to use the old method, with a hand-made polyglot.ini file in wich you specify Polyglot logging should be on:

Code: Select all

Log=true
LogFile=polyglot.log


in the [Polyglot] section.

[Michel]

And one other thing (don't know if it's the fault of Winboard or Polyglot?), that seems to happen only with my own engine ChestUCI (but only in WB+PG):
The loading of ChestUCI takes quite a long time (several seconds), and this also on my faster desktop!?
I have no idea what could cause this, it doesn't happen in other GUIs (e.g. Arena) and also not with other UCI engines.

There are various tests to perform.

Does the delay happen if you start polyglot manually:

polyglot -ec engine.exe

or

polyglot engine.ini

where engine.ini is a PG ini file? In this ini file you can enable logging. The log file contains timestamps which may perhaps indicate where the delay is.

[H.G.Muller]

Here everything happens in about 0.5 sec. Most of the time is spend waiting until WinBoard sends the xboard command.
[deit] beware, I now see this is with a pretty old Polyglot (UCIb22 ?)

Code: Select all

2985427.688 POLYGLOT *** START ***
2985427.688 POLYGLOT INI file "chest.ini"
2985427.688 Adapter->Engine: uci
2985427.798 POLYGLOT *** SETTING BOOK ***
2985427.798 POLYGLOT BOOK "default_book.bin"
2985428.178 GUI->Adapter: xboard
2985428.178 GUI->Adapter: protover 2
2985428.178 Adapter->GUI: feature done=0
2985428.178 Adapter->GUI: feature analyze=1
2985428.178 Adapter->GUI: feature colors=0
2985428.178 Adapter->GUI: feature draw=1
2985428.178 Adapter->GUI: feature ics=1
2985428.178 Adapter->GUI: feature myname="ChestUCI Ver.5.1"
2985428.178 Adapter->GUI: feature name=1
2985428.178 Adapter->GUI: feature pause=0
2985428.178 Adapter->GUI: feature ping=1
2985428.178 Adapter->GUI: feature playother=1
2985428.178 Adapter->GUI: feature reuse=1
2985428.178 Adapter->GUI: feature san=0
2985428.178 Adapter->GUI: feature setboard=1
2985428.178 Adapter->GUI: feature sigint=0
2985428.178 Adapter->GUI: feature sigterm=0
2985428.178 Adapter->GUI: feature time=1
2985428.178 Adapter->GUI: feature usermove=1
2985428.178 Adapter->GUI: feature memory=0
2985428.178 Adapter->GUI: feature smp=0
2985428.178 Adapter->GUI: feature egt=""
2985428.178 Adapter->GUI: feature variants="normal"
2985428.178 Adapter->GUI: feature option="HashSize -spin 64 -1 447"
2985428.178 Adapter->GUI: feature option="DefenderMoves -check 0"
2985428.178 Adapter->GUI: feature option="ProblemType -combo *Mate· /// Stalemate /// SelfMate /// SelfStalemate /// HelpMate /// HelpStalemate"
2985428.178 Adapter->GUI: feature option="SearchDepth -spin 10 0 126"
2985428.178 Adapter->GUI: feature option="AllSolutions -check 1"
2985428.178 Adapter->GUI: feature option="FullMovelist -check 1"
2985428.178 Adapter->GUI: feature option="CompleteDuals -check 0"
2985428.178 Adapter->GUI: feature option="UseDatabase -check 1"
2985428.178 Adapter->GUI: feature option="SearchMode -combo *Normal /// Special /// Automatic /// AutoTurbo /// RetroAnalysis /// ParameterAnalysis /// DatabaseAnalysis /// EgtbAnalysis"
2985428.178 Adapter->GUI: feature option="ChecksOnly -spin 0 -1 31"
2985428.178 Adapter->GUI: feature option="ThreatDepth -spin 0 -125 125"
2985428.178 Adapter->GUI: feature option="ThreatRange -string 3"
2985428.178 Adapter->GUI: feature option="KingSquares -spin 0 -9 9"
2985428.178 Adapter->GUI: feature option="PieceLimit -spin 0 -16 16"
2985428.178 Adapter->GUI: feature option="MaxMoves -spin 0 -222 222"
2985428.178 Adapter->GUI: feature option="SearchOrder -combo *KPX /// KXP /// PKX /// PXK /// XKP /// XPK /// only_CR /// only_CRK"
2985428.178 Adapter->GUI: feature option="RetroOptions -spin 7 0 7"
2985428.178 Adapter->GUI: feature option="EgtbCache -spin 16 1 447"
2985428.178 Adapter->GUI: feature option="EgtbPath -string <empty>"
2985428.178 Adapter->GUI: feature option="ShowAnalysis -check 1"
2985428.178 Adapter->GUI: feature option="ShowProgress -check 1"
2985428.178 Adapter->GUI: feature option="ShowStatistics -check 0"
2985428.178 Adapter->GUI: feature option="EngineMode -combo *UCI /// Chest /// Expert"
2985428.178 Adapter->GUI: feature option="ExpertOptions -string -rS"
2985428.178 Adapter->GUI: feature option="Show Options -button"
2985428.178 Adapter->GUI: feature option="AutoClipboard -check 1"
2985428.178 Adapter->GUI: feature option="AutoLogfile -check 0"
2985428.178 Adapter->GUI: feature option="Show Logfile -button"
2985428.178 Adapter->GUI: feature option="AutoSolution -check 0"
2985428.178 Adapter->GUI: feature option="Show Solution -button"
2985428.178 Adapter->GUI: feature option="TextViewer -string <empty>"
2985428.178 Adapter->GUI: feature option="CpuInfo -string <empty>"
2985428.178 Adapter->GUI: feature option="German -check 1"
2985428.178 Adapter->GUI: feature option="Polyglot Log -check 1"
2985428.178 Adapter->GUI: feature option="Polyglot LogFile -string polyglot.log"
2985428.178 Adapter->GUI: feature option="Polyglot UseNice -check 0"
2985428.178 Adapter->GUI: feature option="Polyglot NiceValue -spin 5 0 20"
2985428.178 Adapter->GUI: feature option="Polyglot Chess960 -check 0"
2985428.178 Adapter->GUI: feature option="Polyglot Resign -check 0"
2985428.178 Adapter->GUI: feature option="Polyglot ResignMoves -spin 3 0 10000"
2985428.178 Adapter->GUI: feature option="Polyglot ResignScore -spin 600 0 10000"
2985428.178 Adapter->GUI: feature option="Polyglot MateScore -spin 10000 0 1000000"
2985428.178 Adapter->GUI: feature option="Polyglot Book -check 1"
2985428.178 Adapter->GUI: feature option="Polyglot BookFile -string default_book.bin"
2985428.178 Adapter->GUI: feature option="Polyglot BookRandom -check 1"
2985428.178 Adapter->GUI: feature option="Polyglot BookLearn -check 0"
2985428.178 Adapter->GUI: feature option="Polyglot KibitzMove -check 0"
2985428.178 Adapter->GUI: feature option="Polyglot KibitzPV -check 0"
2985428.178 Adapter->GUI: feature option="Polyglot KibitzCommand -string tellall"
2985428.178 Adapter->GUI: feature option="Polyglot KibitzDelay -check 0"
2985428.178 Adapter->GUI: feature option="Polyglot ShowPonder -check 0"
2985428.178 Adapter->GUI: feature option="Polyglot UCIVersion -spin 2 1 2"
2985428.178 Adapter->GUI: feature option="Polyglot CanPonder -check 0"
2985428.178 Adapter->GUI: feature option="Polyglot SyncStop -check 0"
2985428.178 Adapter->GUI: feature option="Polyglot RepeatPV -check 0"
2985428.178 Adapter->GUI: feature option="Polyglot PromoteWorkAround -check 0"
2985428.178 Adapter->GUI: feature done=1
2985428.218 GUI->Adapter: accepted done
2985428.228 GUI->Adapter: accepted analyze
2985428.228 GUI->Adapter: accepted colors
2985428.228 GUI->Adapter: accepted draw
2985428.228 GUI->Adapter: accepted ics
2985428.228 GUI->Adapter: accepted myname
2985428.228 GUI->Adapter: accepted name
2985428.228 GUI->Adapter: accepted pause
2985428.228 GUI->Adapter: accepted ping
2985428.228 GUI->Adapter: accepted playother
2985428.228 GUI->Adapter: accepted reuse
2985428.228 GUI->Adapter: accepted san
2985428.228 GUI->Adapter: accepted setboard
2985428.228 GUI->Adapter: accepted sigint
2985428.228 GUI->Adapter: accepted sigterm
2985428.228 GUI->Adapter: accepted time
2985428.228 GUI->Adapter: accepted usermove
2985428.228 GUI->Adapter: accepted memory
2985428.228 GUI->Adapter: accepted smp
2985428.228 GUI->Adapter: accepted egt
2985428.228 GUI->Adapter: accepted variants
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted option
2985428.228 GUI->Adapter: accepted done
2985428.238 GUI->Adapter: new
2985428.238 Adapter->Engine: isready
2985428.238 POLYGLOT NEW GAME
2985428.238 POLYGLOT WAIT
2985428.238 Adapter->Engine: ucinewgame
2985428.238 GUI->Adapter: random
2985428.238 GUI->Adapter: ics -
2985428.238 GUI->Adapter: level 40 1 0
2985428.238 GUI->Adapter: post
2985428.238 GUI->Adapter: hard
2985428.238 POLYGLOT WAIT
2985428.238 GUI->Adapter: easy
2985428.238 POLYGLOT WAIT
2985428.238 GUI->Adapter: ping 1
2985428.238 Adapter->GUI: pong 1
2985428.238 Engine->Adapter: readyok
2985473.754 GUI->Adapter: option HashSize=32
2985473.754 Adapter->Engine: setoption name HashSize value 32
2985473.754 Adapter->Engine: isready
2985473.754 GUI->Adapter: option SearchMode=EgtbAnalysis
2985473.754 Adapter->Engine: setoption name SearchMode value EgtbAnalysis
2985473.754 Adapter->Engine: isready
2985473.754 GUI->Adapter: option Polyglot ResignMoves=5
2985473.754 GUI->Adapter: option Polyglot Resign=1
2985476.798 GUI->Adapter: result * {xboard exit}
2985476.798 POLYGLOT GAME END
2985476.798 POLYGLOT WAIT
2985476.798 GUI->Adapter: force
2985476.798 POLYGLOT WAIT
2985476.798 GUI->Adapter: ping 2
2985476.798 Adapter->GUI: pong 2
2985476.798 GUI->Adapter: quit
2985476.798 POLYGLOT *** "quit" from GUI ***
2985476.798 POLYGLOT *** QUIT ***
2985476.798 Adapter->Engine: quit
2985476.798 POLYGLOT Closing engine
2985476.828 POLYGLOT Calling exit


[F.Huber]

There are various tests to perform.

Well, I've made now different tests (with PG+WB logs), and this delay seems to be just an illusion!
Both logs don't show any timing problems, and now I've also tried to very quickly make a move on the board, and ChestUCI in fact starts immediately.
I only thought it would take so long to load the engine, because the mouse-cursor shows this sandbox-picture for almost 10 seconds after having started WB. But now I see that nevertheless I can start the engine search immediately (even with this sandbox still shown).

Very strange - maybe again only a Win98 problem!?

Or maybe one of you has implemented a virus in PG or WB, which shredders my harddisk in the background?

[F.Huber]

The comment window can be suppressed by the option /autoCommentPopup=false. It is saved in the winboard.ini file, but unfortunately the compiled-in default for this option has always been "true". There is no menu item yet to control this feature from the menus, you would have to give it on the command line (or type it behind your engine name or in the field "additional options" of the startup dialog, or edit it in the winboard.ini.

I've tried it but it doesn't work: "Unrecognized argument /autoCommentPopup"
No success, neither on the commandline nor in the INI-file, and also in Winboard.exe this token doesn't even exist!
There's a 'autoComment' (in EXE and INI), but I guess this only controls the commenting in the history.

Update: Ok, solution found, the correct option's name is /autoDisplayComment.

[H.G.Muller]

Oops, very sorry. I mixed up the name of the option with the name of the function that does it...

[F.Huber]

Oops, very sorry. I mixed up the name of the option with the name of the function that does it...

No problem, I had just a bit to search until I found it in the WB helpfile.

BTW, in both WB helpfiles (RTF, CHM) it is listed as autoDisplayComments, but in WB its name is without the 's'.

[H.G.Muller]

Ah, thank you. We will correct that.