In the saved game, the solver says, the game is unwinnable, but it actually is winnable. Two moves (Col 9 -> Col 10, Col 1 -> Col 10) make the solver correctly report the game as winnable. These two moves are actually reversible (so you could get an infinite loop of solvable / not solvable). I'm attaching the save game (sorry, no clue how I got to that point) and a text file with the steps to actually win the game. Reproducible: Always Steps to Reproduce: 1. Open saved game 2. See "unwinnable" message 3. Use attached solution to win the game Actual Results: Game is winnable Expected Results: Solver should not claim it's unwinnable - at worst it should say "undeterminable", at best it should "know" that the game is winnable
Created attachment 83248 [details] Saved game in "unwinnable" state
Created attachment 83249 [details] The solution to the saved "unwinnable" game
Togehter with a playability upgrade for spider (see Bug report 248859) I also submitted a rewrite of large portions of the spider specific part of the solver, that - according to my tests - fixes this issue. After loading, the game is no longer reported as "unsolvable", but as "undeterminable". After the first suggested move (Col 9 -> Col 10), the game is reported as "solvable".
Sorry for the extremely late comment: when loading the game, Kpatience (3.6, KF 5.33, Qt 5.7.1) on KDE Neon User stable (Plasma 5.9.5) still reports the saved game as unwinnable.
The new solver (417bdc2ec4fa474e3c62063790cae72bc171d7e9) will now say "unable to determine if winnable" as it does not attempt to split the S4 from the S5 (the first move in your solution). Once that move is done, it switches to "is winnable"