How to get InfoDict play result from thinking or pondering engine?

[JohanSjoblom]

Hi all.
I would need to get the InfoDict information from the engine after my call to play() and before the engine returns with a move. In other words; i would like to get InfoDict information on engine thinking while its thinking about a move. I use the asyncio versions of play, analysis and analyse.

The two options I can see from the documentation are indefinite analysis async task (or thread) or a one-time analyse() with for instance a time limit of 0.1sec. BUT: I don't think that tells me anything about the engine thinking. I might be wrong, but I have tested both and they do not seem to give any information on the engine thinking process. These tasks or threads seems to be parallel to the play() call? I might be wrong of course.

I do get information on the thinking process when the engine decides to do its move. But I would like to get the information both while the engine is thinking about the move,... AND also when its the user turn and engine is pondering.

I think this was possible in the old 0.25 python-chess library. I would need the same functionality as I am porting the Picochess software from the old library to this new one. See: https://github.com/JohanSjoblom/picochess

All help appreciated :-)

[JohanSjoblom]

I have a feeling that this cannot be done with this library since InfoHandler is no longer used. I also assume that I cannot run play() and analysis() in the same engine instance (I tried and got async error from library).

If these assumptions are correct I am left with the following options?

1. Dont use Play() to get moves from engine, use analysis() instead. The downside is that I have to implement my own time management and in a way therefore I am not playing against the engines ”time management”. The other downside with this is that old engines probably do not support this analysis?

2. Start another engine instance and run a parallel analysis() there. The downside is CPU usage (my main target is Pi3,4,5). the other downside is that it does not exactly match the thinking of the playing engine instance, especially on NNUE engines. But it might be considered close enough; leaving CPU as the main challenge.

Again, any comments, or even partial help or hints would be appreciated :-)