Oppo Bishops Case Study

OPPOSITE-coloured bishops are notoriously drawish in endings (but not in middlegames).
Sometimes it is possible to hold such an endgame when two pawns down.
At other times, however, even equal pawns do not guarantee a draw.
Often the best way to win such an endgame is to create a passed pawn on each wing, even if that allows the opponent a passer, or passers, elsewhere.

Position after 36 moves in Spanton (1851) - Glenn House (2188), Jersey round two

Material is equal - in numbers, that is, not in quality.
Black's bishop is much more active than White's, and his king is more central. My b2 pawn is en prise and my g5 pawn is vulnerable to ...Bf4, when Black would be simultaneously protecting his e3 passer and tying my rook down.
The only way to free the rook would be to play h2-h4, but then the h pawn would be a new weakness.
I decided my best chance was to activate my rook on the open file, and so played ...
37.Be2 Rd8!?
My analysis engines slightly prefer 37...h4, but the text is their second choice.
38.Rd1 Rxd1+!? 39.Bxd1 Bxb2
Black has won a pawn and left me with two isolated queenside pawns, but I felt I had some reasonable drawing chances.
40.f4 
Activating the king with 40.Kg2 is met by 40...Be5, when ...Bf4 will force White to give up the g5 pawn or play h4, when the king would be tied to the kingside.
40...e2! 41.Bxe2 Bc1 42.f5 Bxg5 43.fxg6 fxg6
Black is 'only' one pawn up, but he has chances to create a passer on both flanks, and White has four isolanis. The game continued ...
44.Kg2 Kf6 45.Kf3 Ke5 46.Bd3 Bc1 47.h3 g5 48.a4 Ba3 49.Ke3 Bc5+ 50.Kd2 Kf4 11mins less 51.Be2 g4 52.hxg4 hxg4
One passer created.
53.Ke1 g3 54.Kf1 Kxe4 55.Kg2 Bd6 56.Bf1 Ke3 57.Kh3 Kd2 58.c5 Bc7 59.Kg2 Kc3 60.Be2 Kb4
I finally resigned.
Doubtless neither of us played perfectly, but I suspect this was an opposite-coloured bishop ending that could not be held.