The diagram position is from a king's gambit decline.

Sepp did not play the paradoxical ...f6 in reply to Bg5. But my computer engine likes the move. I do see the merits black doesn't hinder the development of his light square bishop by playing ...Qd7, which is the actual move played by Sepp. After the possible sequence exf6 gxf6 Bh6, if black can successfully castle queenside he will have excellent play on the kingside. Is there a continuation for white that prevents this plan?

I'm no Kings' Gambit expert, but I would probably just exchange on c6 and the address the issue on g2, maybe just by castling, but I'm not sure it's the safest.

I only know the accepted lines.

i dont like2.f4 at all

At first glance, I'd think 13.O-O might be OK-- get the Rook over to one of the files where it can do some good.

Also, what about 13.Qe2+ ?  If Black doesn't want to lose his right to castle Queenside by moving his King out of check he has to choose between 13...Ne7 or 13...Qe7 14.Qxe7 Nxe7 15.O-O , both of which remove the pesky Knight from d5 (actually, 15.Nc3 might be better, to prevent [Black's] Knight from returning [to d5], and then castle on the next move and now the Rooks are connected and ready to move to those open files).

Edited [for clarity]