Oh Dear

I OFTEN find I can tell how a tournament will go depending on how I perform in round one.
If that turns out to be true in Malta, which began yesterday in the seaside town of Bugibba, I am in for a rough ride.
Spanton (1940) - Marloes Rogge (1442)
Queen's Gambit Declined, Exchange Variation
1.c4 e6 2.Nc3 Nf6 3.d4 d5 4.cxd5 exd5 5.Bg5 c6 6.Qc2 Be7 7.e3 h6

Overwhlemingly most popular here is 8.Bh4

8.Bxf6!?
Not a popular choice, but it has been played by some strong players, including a 2586.
My idea was to divert Black's dark-square bishop from the f8-a3 diagonal so as to speed White's minority attack.
8...Bxf6 9.b4?!
Consistent, but premature. Better first was 9.Bd3.
9...0-0
Not 9...Qe7? 10.Nxd5!
10.Bd3
This loses a pawn, but White's position was already difficult, eg 10.b5 c5!? 11.dxc5 Bf5! 12.Bd3 Bxd3 13.Qd3 Nd7! with excellent play for Black thanks to White's queenside weaknesses.
10...Qe7 11.Nge2 Qxb4
White is a pawn down and has given up the bishop-pair, but has some compensation in the form of a lead in development.
Just five moves later we reached the following position, where I missed a chance to play what seems to be a straight equaliser:

White to equalise with his 17th move

*****
*****
*****
*****
*****
17.g3?
Not-so-hard-to-find is 17.Nb5, especially as White's play has been directed at a queenside coup.
Clearly the reply 17...cxb5 runs into the simple 18.Bxc8.
If 17...Bxf5, then 18.Nxd6 Bxc2 19.Rxc2 c5 20.dxc5 is fine for White.
That leaves 17...Qe7?! 18.Bxc8 Rxc8 (or 18...cxb5 19.Qc6 Rb8 20.Nxd5 Nxd5 21.Qxd5 with much the better game for White) 19.Qf5, when Black is in trouble, eg 19...Qd7? 20.Qxd7 Nxd7 21.Nxa7 Ra8 22.Nxc6 Rxa2 23.Nxd5, when it is White, not Black, who is a pawn up.
Despite this miss, I reached the following late-middlegame where White still has a fair bit of compensation for the pawn-minus:

White has just played 31.Rb2-b4

31...b5?
Black had to play 31...Rc6, when my main analysis engines, Stockfish9 and Komodo9, like 32. Rc1. Black is still better, but White has decent drawing chances.
32.Rxc4?
White equalises the material, but would have been a pawn up after 32.axb5 axb5 33.Rxc4, eg 33...bxc4 34.Rxb8+ Kh7 35.Rc8, etc.
32...Nd5 33.Nxd5 Rxd5 34.axb5
Better was 34.Rc6! a5 and now the move I missed, 35.e4, eg 35...Rh5 36.Kg4 Rxh2 37.axb5 (analysis by Stockfish9 & Komodo9), when Black's rooks are discoordinated while White has a fast-moving passed pawn.
34...Rdxb5 35.Ra1 a5 36.Rc7 Ra8
Stockfish9 reckons the position is dead-equal; Komodo9 slightly prefers White. I manged to get down to a position of rook and two pawns versus rook-and-one, but the simplification meant White could not progress further.