lichess.org/practice

Hi, I know it's still in development but I thought it might be useful to have a full list of the bugs (at least the ones that I found).

Checkmating Patterns II - Balestra Mate #1 -- Puzzle appears to be just an analysis board
gyazo.com/820a376e6e40a314e53a839221bdf5e5
gyazo.com/8979fc3ab9f039dc31f639091d57a92d

Checkmating Patterns IV - Kill Box Mate #1 -- Puzzle appears to be just an analysis board
gyazo.com/32a9bb8b0dd5b536aaa4bdc3c37de0ae
gyazo.com/6bde260e2cc46b50688b0effddc04f22

The Fork - Fork Challenge #7 P -- ends in failure after following computer recommended line gyazo.com/c699308ce0254191a96199cdc64c93f8

The entire discovered attack portion was completely bugged for me. A few of the bugs were: the computer thinking endlessly, puzzles being failed after following the best line but succeeding on the second attempt in the same line, and puzzles endlessly failing following the computer suggested line.
All of the discovery puzzles were fixed by refreshing the page except for Discovery Attack #4 which continued to fail the correct solution.

Overloaded Pieces - Overloaded #1 -- Correct solution fails the puzzle gyazo.com/66e26e548bbd0360791bbcb0739f87ab

Zwischenzug - Zwischenzuq Challenge #2 -- Correct solution fails the puzzle gyazo.com/4b3c59d635a4f78bbc3fb7862bf21ae2?token=973879a5155008a86946800aff076e15

Zwischenzug - Zwischenzuq Challenge #4 -- Puzzle appears to be just an analysis board

Side Note: The two knight checkmate is very very difficult. So difficult that the computer suggested line did not lead to checkmate after 50 moves. If you wish to include this, I suggest using the tablebase for suggestions or making it a simpler puzzle with the king trapped in the corner.

Side Side Note: The first zugzwang puzzle is a confusing example of the theme. It is technically zugzwang because black is forced to move his king away from attacking the rook on d6 after the first move, but the puzzle seems to be more focused on the theme of a trapped piece.

Side Side Side Note: As the guy who likes to pointlessly speedrun things, I would love it if all of the practice puzzles that run using stockfish used either a set solution (in the style of lichess.org/training) or a tablebase as waiting for the local stockfish to load is tedious and time consuming.

Thank you very much for your continual work on improving lichess and adding features!

@TheLlamaLord
Sir, your speedrunning is not "pointless" it's beautiful. Please let me know when the /practice speedrunning will take place so I can call in sick to work, get a babysitter for my kid, divorce my wife etc.

Thanks @TheLlamaLord we will fix all that.

We're going to cache the computer evaluations, so if you play the same line again, it will be near instantaneous.

Thanks for the feedback @TheLlamaLord, you are the hero we need!

I have fixed all the puzzles that didn't start and I have also fixed the puzzles where the correct solution was not recognized. It's not trivial to make deterministic puzzles when the engine is involved, but it should be mostly okay now.

Response to Side Note: As stated in the puzzle, checkmate is possible as soon as 15 moves. If the engine sees this mate in 15, it should give you the correct answer. I do think you are correct in saying that we don't give hints based on tablebases at this moment so I added it to our backlog, but can not promise if or when it will be implemented.

Response to Side Side Note: How would you illustrate it?

Response to Side Side Side Note: You should already be noticing big performance improvements, and even better in the coming days.

Discovered attack #5 consistently fails the correct line

The Fork Challenge #7 P Fails about 80% of the time against the correct line

Discovered attack #4 still fails about half of the time against the correct line

The rest appears to be fixed!

I don't know how I would like the first zugzwang puzzle to be presented, but I just feel that the other four puzzles do a lot better at demonstrating the idea behind the theme.

Thanks for all your hard work!

@TheLlamaLord The three remaining problems should be fixed now. Thanks for valuable feedback.

Unfortunately, I think Fork Challenge #7 is just an analysis board

Discovery Attack #5 works when the exact computer line is played, but variations with a similar stockfish evaluation are failed occasionally.

Discovery Attack #4 does appear to be fixed :)

Note that you can copy the URL to a specific practice chapter, for instance en.lichess.org/practice/tactics/discovered-attacks/MnsJEWnI/n76eNBSa

Is practice still in development or is there just no link to it?

TheLLamaLord is awesome.