Leon P Smith

IIRC, the only thing SCRAM-SHA-256-PLUS does is specify a larger minimum number of iterations in the PBKDF2 password hash function. So it is utterly trivial to implement once you have...

Ok, n/m, I was completely wrong. I am reading up on SCRAM at the moment because I need some deeper knowledge of it to inform some preliminary choices in a...

There is a dependency between `getNotification` and a higher-level binding: in hasql's and postgresql-simple's case, it is the `MVar`. Your update has drastically changed the concurrent semantics for the worse....

Oh, no apology needed regarding the time taken, I am certainly in no rush. (Though I don't speak for begriffs) As for the "generic" getNotification, that's not an interface that...

Yes, that function should be sufficient for my purposes.

However, while I'm 100% ok with such a solution, the "generalized" `getNotification` that Nikita proposed still isn't very general: it's assuming that it gives you exclusive access to a libpq...

The code is BSD3 licensed in postgresql-simple, so you already have permission. If you want explicit permission otherwise, by all means, please. :)

I would be extremely interested in looking at the test case. Cheers!

Well, I'm not surprised that there is a race condition; you removed the code to prevent several race conditions!

Ok, so the `join $ withLock fooLock $ do {...}` and the `join $ atomically $ do {...}` idioms are quite useful when dealing with concurrent programming; the idiom allows...