As many use this feature, I decided to postpone working on Alekhine's games and instead spend a week upgrading the Analysis Board. The old code bad to say the least, so I basically redesigned everything from scratch. I made a new UI entirely:
New features:
Multiple engine lines are now Supported:
Edit:
As of now, Stockfish 17.1 is now bundled Chess Tools. I previously didn't do it for legal reasons because you need to make your source code public under their license. However if you simply link the exe externally and don't modify the code, you can simply link to where to get Stockfish and not have to make your whole game source code open source.
So in analysis you have the option to use Stockfish 17.1 or your custom engine of choice, say if you want to use Leela or something. In Play mode, you will still go against my engine but I might make stockfish an option in the future too.
Game Review
This is similar to before but more improved. It should be a lot more reliable now.
The review will use the engine you have currently chosen, either default or custom. Then you Run Review and wait for it to finish. After you will see the score graph and a count of inaccuracies and blunders.
Icons will now be shown on the board, as well as in the log. You can disable this in the review section.
Currently only inaccuracies and blunders are shown, as they are a lot simpler to implement.
Full customisable blunders and inaccuracies
Instead of just hard coding the values, I tried to make them as customisable as possible:
-Inaccuracies go from 0.3 to 1.5 points
-Blunders go from 1 to 3.5 points
So if you want it to be lenient and not nit-pick every possible drop in score, you can do that.
And if you want it to criticise everything, like a masochist, then you can do that also.
Still Winning Threshold
Another setting is the still winning threshold. This is basically for when it shouldn't matter if you play a slightly inaccurate move. This goes from 3, 4, 5, 6 or none.
Basically how this works is that if you are winning by 5 points or more, then it generally doesn't matter if you didn't take a pawn or not. You're still winning if your score is above this threshold, so no blunders or inaccuracies will be marked. Set this to none to turn this off.
Manual Review
This is an entirely new idea. As well as automated review, you can turn on the engine and have it save scores for each move you look at. Blunders and inaccuracies will then be highlighted automatically. The score will be saved in the log. This can also overwrite an automatic reviews score.
Let's say that the engine said something was a blunder and you are doubtful, well you can simply turn on the engine and look at the position again. If the score was inaccurate, then the new score will overwrite it. You can turn this feature off in review settings too.
If you want the counts and graph to be updated then simply click Update Info after the manual review.
Variations
To create a variation, you simply play a move at any point except for on the last move played.
Before I would insert variations into the same log section. This was honestly a complete nightmare to get working and caused a lot of headaches. Such as having to move every other element after the new along by one.
I decided to remove that entirely and now I have a separate box for variations. The way this works is that each variation is linked to it's starting move.
For example you create a variation from [2.Nf3]. This variation will only be visible when that move is selected in the main window. If you click to another move, then those other variations will be shown instead, if they exist.
I only allow one variation per main move and you can't stack them, for simplicity. If you play another move when not in the last position while there is already a variation for that base move, then the previous variation is cleared and the new takes its place.
Still to do:
-Improve the default engine and make it support multiple principle variations, meaning show multiple best moves in a position.
-Show best moves and only moves. Maybe brilliant moves too but that algorithm is far from simple.
-When you make a new variation, it doesn't actually check if the next move was the same move. I plan to fix this in the future.
-My long term goal is to have something that judges the complexity of a position based on the amount of good vs only moves available. Instead of just saying "Blunder!", it could say: "Blunder but there was only one good move." or instead: "Blunder! and there were many good options".
Fabiano Caruana game review:
Added new commentary, moves to find and variations to Caruana's games.
Before moves to find: 147
Now moves to find: 306