Here is a position that was shown to me by a fellow club member (John Leitel),
And the question is who wins, because white move the bishop to f3 and called checkmate, but then black said no it is not and in fact you are checkmated after I move the pawn to d5. White again said no you are wrong because I capture that pawn enpassant and therefore the pawn never made it to d5 and you were still checkmated.
En passant is an optional response to a pawn move.
In this position it is obviously black’s move, since black is in check.
d5 is a legal move, and it also puts white in check. Because of that check, the pawn on e5 cannot capture en passant on d6, since that does not block the check.
So black wins.
This position (or a similar one) appeared in a book by Assiac (which is Caissa spelled backwards) a few decades ago. It was a fun book.
The author’s point was that, if you consider the spirit of the en passant rule (and if you note that en passant means “in passing”), then black’s d7 pawn never reached d5. It was captured at d6 when it tried to pass through. Therefore, it never blocked white’s check. “Dead men cannot shoot.”
The author concluded that, although every chess player would surely agree with the literal interpretation (as explained by Nolan above), nevertheless there was something just a little disturbing about the whole thing.
As I said, it was a fun book (I think it was called The Pleasures of Chess).
Bill Smythe
On re-reading the first post, I have to edit to clarify myself.
If the light-square Bishop of Black’s had been on f7 instead of e8, White wins.
(…d5 exd6+ Bd5 Bxd5#, a move longer than …Bd5 Bxd5#.)
(Then again, with the Bishop on f7, there was never any check to White… and no counter-checkmating move possible. 
)
But, as Montgomery Scott said, “Aye… and if my Grandmother had wheels, she’d be a wagon!” 
It’s not disturbing when one considers that, even if the pawn were “sliding through” the 3rd/6th ranks, it hasn’t yet passanted the other pawn. (It ain’t “passing” the other pawn until it arrives alongside… ) But it is a little wiggy to consider that a pawn moving two squares isn’t actually sliding through that intermediate rank, yet in Castling the King certainly does slide through the intermediate file(s) - hence you can’t castle through a check.
Maybe someday there would be a special rule for taking a pawn en réunion, and then the notion would be correct.
This, however, is incorrect because as soon as black moves his d pawn toward d5, white no longer has him in check, since white can no longer capture black’s king on the next move. The reason is simple: If white captures the d pawn as it passes, he has made his move, and therefore cannot capture black’s king until the following move. But if white does not capture the d pawn as it passes, it ends up on the d5 square blocking white’s bishop attack on black’s king, and white cannot capture black’s king.
Actually, of course, white will not be able to capture the d pawn en passant, since doing so would guarantee black another move which black could use to capture white’s king via the bishop attack that was revealed as soon as black advanced the d pawn.
Bob