Does anyone have any experience in this line?

It runs 1. e4 e6 2. d4 d5 3. Nc3 Nf6 4. Bg5 Be7 (I think this might already not be too good for black) 5. e5 Nfd7 6. h4! sacrificing a pawn for initiative and a powerful half-open h-file.

Here's a miniature I just finished after coming across this line for the first time. I have to say it was very natural to play, and a welcome change to the normal games you get with the French.

Your mini is nice.  Question, and maybe i'm overlooking this, but why did you rule out Nxd6 on move 12?

Very nice, very enjoyable miniature too... gimly had an Interesting question about Nxd6... it does look strong... I disagree with your annotation that the fault in 13.c5? is that it neglects development -- that comment is both generic and inaccurate... no James Joyce fan should resort to lazy cliche, Gabriel -- I think 13.c5 is a stately, plump, tactical blunder that allows the crushing Nb5.

I was rather under the impression that Black shouldn't take the pawn.

You're right that 12. Nxe6 was strong. However, after 12...g6 13. Nxc7+ Kd8 14. Qg5+ Kxc7 15. Nb5+, I couldn't see a definite way to continue the iniative, so I stuck with 12. Ng6.

marvellosity, what would you suggest instead of taking the pawn (I think you're right)? Just to continue with typical French plans (c5, and so on)? Black certainly can't castle kingside.

i think i have seen from a 175 player was a sort of winerwa variation after bg5 bb4 and best is to block with the bishop

I'm not sure what you mean. After 4...Bb4, I'm pretty sure the best move is 5. e5 (5...h6).

Nice miniature, but Black doesn't have to play that bad. Instead of 9. ... f6 - g6, and then Nc6, Nb6, Bd7 and O-O-O. Max White can get on the kingside after that is it's pawn back and Black has good perspectives on the queenside. If White castles queenside c5 could be played after Kb8 and Rc8. This plan is almost a refutation of Alekhine-Chatard attack from my point of view. This is from one of my games, I am Black, White's 22nd move was actually worse than Qd2: