One thing I really like about the Simon Tatham version is that you never have to guess. I think that would make a great addition to your rule changes, because it would mean you always die if you mistakenly open a square that isn't safe. On the ST version, since you're still allowed to guess, there's still a luck element, but this would close that loophole. On the other hand, it would also mean you don't spend an endless amount of time trying to figure out if you have to guess (in your version), because you're guaranteed that you don't.
Incidentally, I think this change would possibly speed up the game on low power devices (it's quite slow on my phone at present). When this setting is enabled, you can generate a single static puzzle at the beginning of the game: the only time you need to regenerate the puzzle is when the user guesses, and then you generate a puzzle that instantly kills them. (Of course, I'm assuming most of the power use is in generating puzzles, not in backtracking to find safe squares when the user opens a new square. That might not be true.)
One thing I really like about the Simon Tatham version is that you never have to guess. I think that would make a great addition to your rule changes, because it would mean you always die if you mistakenly open a square that isn't safe. On the ST version, since you're still allowed to guess, there's still a luck element, but this would close that loophole. On the other hand, it would also mean you don't spend an endless amount of time trying to figure out if you have to guess (in your version), because you're guaranteed that you don't.
Incidentally, I think this change would possibly speed up the game on low power devices (it's quite slow on my phone at present). When this setting is enabled, you can generate a single static puzzle at the beginning of the game: the only time you need to regenerate the puzzle is when the user guesses, and then you generate a puzzle that instantly kills them. (Of course, I'm assuming most of the power use is in generating puzzles, not in backtracking to find safe squares when the user opens a new square. That might not be true.)