Dear John, the background is a binary format assembler, so you can think of the monad as the Put monad, only sufficiently lazy to admit a useful MonadFix instance. Then one can do nice things like mdo putWord32 offset putBS someMoreHeaderData ... offset <- getCurrentOffset putBS byteString1 where I conceptually use the offset before it is known. So what if I want to put two offsets, followed by two bytestrings? Easy: mdo putWord32 offset1 putWord32 offset2 putBS someMoreHeaderData ... offset1 <- getCurrentOffset putBS byteString1 offset2 <- getCurrentOffset putBS byteString2 Now I try to generalize that to a list of Bytestrings, and just from the looks of it, this is what you want to do: mdo mapM_ putWord32 offsets putBS someMoreHeaderData ... offsets <- forM byteStrings $ \b -> offset <- getCurrentOffset putBS b return offset but that will <<loop>>. Am Mittwoch, den 15.01.2014, 13:10 -0800 schrieb John Lato:
Hello Joachim, I don't really understand what you're doing here. There's the obvious
mapM_ (act2 >=> act1)
But presumably act1 performs some monadic action that doesn't depend on its input, and you need that to be performed before act2?
Exactly, (act2 >=> act1) would write out the data in the wrong order.
To me, it feels like there's some sort of implicit coupling between act1 and act2, and you would be better off extracting that, perhaps by changing act1 to have the type act1 :: M T -> () .
If that's not possible, your approach seems pretty simple to me.
I don’t think its, possible with that signature, no... (It were if I were to interleave the calls to act1 and act2, instead of requiring first all act1 and then all act2, but then it would be trivial anyways.) Greetings, Joachi -- Joachim Breitner e-Mail: mail@joachim-breitner.de Homepage: http://www.joachim-breitner.de Jabber-ID: nomeata@joachim-breitner.de