Crafty 23.1 (JA) oddities

[Dirt]

Note that WinBoard has options -firstScoreAbs true|false / -secondScoreAbs true|false to correct the scores of such non-compliant engines, which can also be set through the "Options -> Adjudications..." menu dialog.

I have tried to use those options is the distant past to make other engines behave like Crafty, but it was unsatisfactory. I'm not sure why now, perhaps it didn't correct the scores to be from white's point of view when in analysis mode.

[hgm]

Non-white-POV programs should be considered non-compliant, for the same reason non-white-POV chess diagrams are non-compliant.

Well, WB supports those too: this is called flipView.

But seriously, what is compliant or not is defined by the standard, not your personal preference. You remind you of my friend, who, after hearing a traffic warning on his car radio that someone on route 69 was driving against the traffic in the wrong lane, said: "One??? I have seen thousands!"

[hgm]

I have tried to use those options is the distant past to make other engines behave like Crafty, but it was unsatisfactory. I'm not sure why now, perhaps it didn't correct the scores to be from white's point of view when in analysis mode.

Good point! I looked at the code, and indeed the score is only flipped with this option in MachineBlack and for the black engine in TwoMachines mode.

I guess this should be fixed: as the engine is not really playing any side, I guess the score presented to the user in analysis mode wshould always be from the white POV, even when black is to move. So WinBoard should flip the engine score (in addition to whatever the ScoreIsAbs option says) in Analysis mode when black is to move, before presenting it to the user in the message field or EngineOutput window.

This comes just in time to change this in the upcoming WinBoard release (4.4.2).

[michiguel]

I have tried to use those options is the distant past to make other engines behave like Crafty, but it was unsatisfactory. I'm not sure why now, perhaps it didn't correct the scores to be from white's point of view when in analysis mode.

Good point! I looked at the code, and indeed the score is only flipped with this option in MachineBlack and for the black engine in TwoMachines mode.

I guess this should be fixed: as the engine is not really playing any side, I guess the score presented to the user in analysis mode wshould always be from the white POV, even when black is to move. So WinBoard should flip the engine score (in addition to whatever the ScoreIsAbs option says) in Analysis mode when black is to move, before presenting it to the user in the message field or EngineOutput window.

This comes just in time to change this in the upcoming WinBoard release (4.4.2).

It is not relevant whether the engine is playing a particular side, scores are reported for the side to move, regardless if there is an engine or not. Please do not break compatibility with the EPD standard!

Miguel

[Michel]

It is not relevant whether the engine is playing a particular side, scores are reported for the side to move, regardless if there is an engine or not. Please do not break compatibility with the EPD standard!

I agree. Reporting the score for the side to move seems very natural even in analysis mode. I would be quite shocked to see this change.

[hgm]

OK, so the score in analysis mode has also be reported from the POV of the side to move. But that still means someting needs fixing: the ScoreIsAbs options now only are active when the engine is playing black in local mode. For sure this should be extended to the case where the engine is playing black on ICS in zippy mode. And I guess it should be extended to Analysis and Analyze File mode as well: when an engine gives absolute (= white POV) scores that we want to see flipped, it will also do it in those modes.

[Don]

Might be an issue with winboard perhaps. Or with the compiler. But not Crafty.

No issue is with me

I changed it to computer's point of view for myself (I prefer it),
but left the changes in to release compile by mistake.

Correct version will be replaced shortly.

My apologies,

Jim.

Is there actually a crafty UCI build? I don't see it on your site but I would like to have it.

[Jim Ablett]

Might be an issue with winboard perhaps. Or with the compiler. But not Crafty.

No issue is with me

I changed it to computer's point of view for myself (I prefer it),
but left the changes in to release compile by mistake.

Correct version will be replaced shortly.

My apologies,

Jim.

Is there actually a crafty UCI build? I don't see it on your site but I would like to have it.

Hi Don,

No, but wb2uci adapter is included.

Jim.

[Don]

Might be an issue with winboard perhaps. Or with the compiler. But not Crafty.

No issue is with me

I changed it to computer's point of view for myself (I prefer it),
but left the changes in to release compile by mistake.

Correct version will be replaced shortly.

My apologies,

Jim.

Is there actually a crafty UCI build? I don't see it on your site but I would like to have it.

Hi Don,

No, but wb2uci adapter is included.

Jim.

That's no good. I'm on linux and the adapter sort of works via wine, but it does not run a 64 bit linux version of crafty. I didn't try a 32 bit crafty but I don't want to run 32 bit binaries - it's a waste of resources.

I don't know if the wb adaptor would even by able to handle a linux binary of any sort. It doesn't seem likely to me that a windows program, even running on an emulator would be able fork off a linux binary.

I may just have to build by own native adapter for Linux. Maybe an excuse to learn the go language by Ken Thompson.

[hgm]

Why not simply run the WinBoard build?

And now that we are discussing this: is there already a Crafty build that can be configured completely through the WinBoard interface, understanding the memory and cores commands, and defining option features for all the options normally in its ini file? That would be much more interesting than a UCI version.