Thank you (and the others) for tips. I'll check the suggestions but yes sequence requires that all items have the same type so that sequence2 would work better but feels un-idiomatic. I needed such code several times in several contexts, so i assumed the answer would be obvious. Maybe i'm structuring my code the wrong way for haskell, i'll rethink those functions. Thanks for now! Emmanuel On Sep 6, 2013 10:37 PM, "Nicholas Vanderweit" wrote:

It seems like he wants something like sequence, but for tuples. Say:

sequence2 :: Monad m => (m a, m b) -> m (a, b) sequence2 (ma, mb) = ma >>= \a -> mb >>= \b -> return (a, b)

But without knowing the use case it's hard to know whether or not this could be done simply with "sequence" and mapM*

Nick

On Fri, Sep 6, 2013 at 2:28 PM, Michael Steele wrote:

...

The `sequence`, `mapM`, and `mapM_` functions may be what you are looking for.

let allTogether = sequence [doA, doB, doC, doZ] case allTogether of Nothing -> ... Just as -> ...

Use `maybe` and `fromMaybe` to avoid case statements:

fromMaybe "failure!" $ do as <- sequence [doA, doB, doC, doZ] return $ "Success: " ++ show (length as)

The `catMaybes` function can be used to execute each computation in the list even if some of them return nothing.

On Fri, Sep 6, 2013 at 12:49 PM, Emmanuel Touzery wrote:

...

Hello,

I'm often using the Maybe monad to combine Maybe computations one after the other without conditionals.

But I'm not sure how to do it, when I have several operations which return Maybe and I want all the indiviual values in the end, not a combination of the values.

Currently I do:

let allTogether = do ma <- doA mb <- doB return (ma, mb) case allTogether of Nothing -> ... Just (a, b) -> ...

This way in this case I have one case instead of two, however I'm sure there must be a nicer way to achieve this result?

Thank you!

Emmanuel

_______________________________________________ Beginners mailing list Beginners@haskell.org http://www.haskell.org/mailman/listinfo/beginners

-- -- Michael Steele

_______________________________________________ Beginners mailing list Beginners@haskell.org http://www.haskell.org/mailman/listinfo/beginners

_______________________________________________ Beginners mailing list Beginners@haskell.org http://www.haskell.org/mailman/listinfo/beginners

Hi, In that case you can use <$> and <*>. Your code will look like, .... <$> doA <*> doB where ... is the rest of the computation Thanks Divyanshu Ranjan On Sat, Sep 7, 2013 at 2:34 AM, Emmanuel Touzery wrote:

Hmm I've been looking some more. Turns out I'm pretty sure... What I've been looking for is liftMx.

In this case, liftM2.

liftM2 ... doA doB

where "..." represents the rest of the computation as it did in the example snippet I sent in my first email.

In the case of the Maybe monad, "..." will be called with the "Just" values and only if both doA and doB are not Nothing.

and for "n" operations, there is "ap", I'll look at that one too.

emmanuel

On Fri, Sep 6, 2013 at 10:43 PM, Emmanuel Touzery wrote:

...

Thank you (and the others) for tips. I'll check the suggestions but yes sequence requires that all items have the same type so that sequence2 would work better but feels un-idiomatic. I needed such code several times in several contexts, so i assumed the answer would be obvious. Maybe i'm structuring my code the wrong way for haskell, i'll rethink those functions. Thanks for now!

Emmanuel

On Sep 6, 2013 10:37 PM, "Nicholas Vanderweit" wrote:

...

It seems like he wants something like sequence, but for tuples. Say:

sequence2 :: Monad m => (m a, m b) -> m (a, b) sequence2 (ma, mb) = ma >>= \a -> mb >>= \b -> return (a, b)

But without knowing the use case it's hard to know whether or not this could be done simply with "sequence" and mapM*

Nick

On Fri, Sep 6, 2013 at 2:28 PM, Michael Steele wrote:

...

The `sequence`, `mapM`, and `mapM_` functions may be what you are looking for.

let allTogether = sequence [doA, doB, doC, doZ] case allTogether of Nothing -> ... Just as -> ...

Use `maybe` and `fromMaybe` to avoid case statements:

fromMaybe "failure!" $ do as <- sequence [doA, doB, doC, doZ] return $ "Success: " ++ show (length as)

The `catMaybes` function can be used to execute each computation in the list even if some of them return nothing.

On Fri, Sep 6, 2013 at 12:49 PM, Emmanuel Touzery wrote:

...

Hello,

I'm often using the Maybe monad to combine Maybe computations one after the other without conditionals.

But I'm not sure how to do it, when I have several operations which return Maybe and I want all the indiviual values in the end, not a combination of the values.

Currently I do:

let allTogether = do ma <- doA mb <- doB return (ma, mb) case allTogether of Nothing -> ... Just (a, b) -> ...

This way in this case I have one case instead of two, however I'm sure there must be a nicer way to achieve this result?

Thank you!

Emmanuel

_______________________________________________ Beginners mailing list Beginners@haskell.org http://www.haskell.org/mailman/listinfo/beginners

-- -- Michael Steele

_______________________________________________ Beginners mailing list Beginners@haskell.org http://www.haskell.org/mailman/listinfo/beginners

_______________________________________________ Beginners mailing list Beginners@haskell.org http://www.haskell.org/mailman/listinfo/beginners

_______________________________________________ Beginners mailing list Beginners@haskell.org http://www.haskell.org/mailman/listinfo/beginners

Hello, And I tried to apply it to the haskell fay compiler, but failed: http://stackoverflow.com/questions/18667530/dont-understand-this-liftm2-beha... Not sure whether it's a fay bug or something about the fay monad or liftM2 which i don't understand. emmanuel On Sep 6, 2013 11:04 PM, "Emmanuel Touzery" wrote:

Hmm I've been looking some more. Turns out I'm pretty sure... What I've been looking for is liftMx.

In this case, liftM2.

liftM2 ... doA doB

where "..." represents the rest of the computation as it did in the example snippet I sent in my first email.

In the case of the Maybe monad, "..." will be called with the "Just" values and only if both doA and doB are not Nothing.

and for "n" operations, there is "ap", I'll look at that one too.

emmanuel

On Fri, Sep 6, 2013 at 10:43 PM, Emmanuel Touzery wrote:

...

Thank you (and the others) for tips. I'll check the suggestions but yes sequence requires that all items have the same type so that sequence2 would work better but feels un-idiomatic. I needed such code several times in several contexts, so i assumed the answer would be obvious. Maybe i'm structuring my code the wrong way for haskell, i'll rethink those functions. Thanks for now!

Emmanuel On Sep 6, 2013 10:37 PM, "Nicholas Vanderweit" wrote:

...

It seems like he wants something like sequence, but for tuples. Say:

sequence2 :: Monad m => (m a, m b) -> m (a, b) sequence2 (ma, mb) = ma >>= \a -> mb >>= \b -> return (a, b)

But without knowing the use case it's hard to know whether or not this could be done simply with "sequence" and mapM*

Nick

On Fri, Sep 6, 2013 at 2:28 PM, Michael Steele wrote:

...

The `sequence`, `mapM`, and `mapM_` functions may be what you are looking for.

let allTogether = sequence [doA, doB, doC, doZ] case allTogether of Nothing -> ... Just as -> ...

Use `maybe` and `fromMaybe` to avoid case statements:

fromMaybe "failure!" $ do as <- sequence [doA, doB, doC, doZ] return $ "Success: " ++ show (length as)

The `catMaybes` function can be used to execute each computation in the list even if some of them return nothing.

On Fri, Sep 6, 2013 at 12:49 PM, Emmanuel Touzery wrote:

...

Hello,

I'm often using the Maybe monad to combine Maybe computations one after the other without conditionals.

But I'm not sure how to do it, when I have several operations which return Maybe and I want all the indiviual values in the end, not a combination of the values.

Currently I do:

let allTogether = do ma <- doA mb <- doB return (ma, mb) case allTogether of Nothing -> ... Just (a, b) -> ...

This way in this case I have one case instead of two, however I'm sure there must be a nicer way to achieve this result?

Thank you!

Emmanuel

_______________________________________________ Beginners mailing list Beginners@haskell.org http://www.haskell.org/mailman/listinfo/beginners

-- -- Michael Steele

_______________________________________________ Beginners mailing list Beginners@haskell.org http://www.haskell.org/mailman/listinfo/beginners

_______________________________________________ Beginners mailing list Beginners@haskell.org http://www.haskell.org/mailman/listinfo/beginners

Well I'm trying to combine two monads here, Maybe and either the Fay or the IO monad. In the example code I just put return but otherwise in my real code "operation" has side-effects and needs the Fay monad in the case of the Fay program. I understand how to use lift if "operation" is pure, without using monads, but I have this trouble when it operates in another monad. I thought I understood it, but looking at the difference in behaviour in the examples I posted, maybe I don't. Emmanuel On Sat, Sep 7, 2013 at 9:11 AM, Daniel Trstenjak

wrote:

Hello Emmanuel,

And I tried to apply it to the haskell fay compiler, but failed:

http://stackoverflow.com/questions/18667530/dont-understand-this-liftm2-beha...

Not sure whether it's a fay bug or something about the fay monad or liftM2 which i don't understand.

The point of lifting is, that you lift an operation into an other "context", so that the operation itself hasn't to operate in the same "context".

I can't see how even your GHC version of 'operation' should work, because in both cases it should be something like:

liftM2 (++) getValue1 getValue2

or in applicative style

(++) <$> getValue1 <*> getValue2

Greetings, Daniel

_______________________________________________ Beginners mailing list Beginners@haskell.org http://www.haskell.org/mailman/listinfo/beginners

I got an helpful answer on stackoverflow that probably explains it all. Otherwise if i add the type, i'll have to bind it to a concrete monad due to the lack of typeclasses, that's why i didn't put a type. Thanks again. On Sep 7, 2013 9:50 AM, "Daniel Trstenjak" wrote:

...

I thought I understood it, but looking at the difference in behaviour in the examples I posted, maybe I don't.

I would try to add a type signature for liftM2, perhaps you're than getting a more helpful error message.

Greetings, Daniel _______________________________________________ Beginners mailing list Beginners@haskell.org http://www.haskell.org/mailman/listinfo/beginners

I can't see how even your GHC version of 'operation' should work ...

Ok, the GHC version puts the IO action into the Maybe and returns it then from 'process'. There's really no need for this extra wrapping of the string into the IO monad. I know nothing about Fay, so I don't know why this extra wrapping shouldn't work for Fay. Fay has an instance for Monad, right? Greetings, Daniel

I know nothing about Fay, so I don't know why this extra wrapping shouldn't work for Fay. Fay has an instance for Monad, right?

well when you write haskell code that you'll compile with Fay you can't use typeclasses. So AFAIK no, the "Fay" that I use in the Fay code doesn't implement the Monad typeclass. OK, so it seems my Haskell code is OK, and probably my stackoverflow question is legitimate about whether this valid Haskell code can work on Fay. Btw I used stackoverflow because the Fay website states that they use it as a preferred channel for support. Thank you for the insights! Emmanuel