($) and ApplicativeDo - Haskell-Cafe - Haskell.org

newer
Trying to test natural...

($) and ApplicativeDo

older
Testing Msgpack system

amindfv＠gmail.com

16 Jun 2016 16 Jun '16

12:08 p.m.

foo :: Applicative f => f String foo = do x <- pure "this works" pure x ... but replace "pure x" with "pure $ x" and it doesn't typecheck: a monad instance is required! Tom

Reply

Sign in to reply online Use email software

Show replies by date

KC

16 Jun 16 Jun

12:24 p.m.

Think of the types of pure x And pure $ x -- -- Sent from an expensive device which will be obsolete in a few months! :D Casey On Jun 16, 2016 9:06 AM, wrote:

foo :: Applicative f => f String foo = do x <- pure "this works" pure x

... but replace "pure x" with "pure $ x" and it doesn't typecheck: a monad instance is required!

Tom _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe

Reply

Sign in to reply online Use email software

amindfv＠gmail.com

12:30 p.m.

They're the same:

:t \x -> pure x \x -> pure x :: Applicative f => a -> f a :t \x -> pure $ x \x -> pure $ x :: Applicative f => a -> f a

Tom

El 16 jun 2016, a las 12:24, KC escribió:

Think of the types of

pure x

And

pure $ x

-- --

Sent from an expensive device which will be obsolete in a few months! :D

Casey

...
On Jun 16, 2016 9:06 AM, wrote: foo :: Applicative f => f String foo = do x <- pure "this works" pure x

... but replace "pure x" with "pure $ x" and it doesn't typecheck: a monad instance is required!

Tom _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe

Reply

Sign in to reply online Use email software

Joe Quinn

12:39 p.m.

I expect it's that the root of the expression on the last line is ($) and not pure. ApplicativeDo has restrictions on what you can do to avoid allowing blocks that have to translate in terms of join/(>>=), and likely one of them is that the last line can't be an arbitrary thing of the right type. On 6/16/2016 12:30 PM, amindfv@gmail.com wrote:

They're the same:

...
:t \x -> pure x \x -> pure x :: Applicative f => a -> f a :t \x -> pure $ x \x -> pure $ x :: Applicative f => a -> f a

Tom

El 16 jun 2016, a las 12:24, KC mailto:kc1956@gmail.com> escribió:

...
Think of the types of

pure x

And

pure $ x

-- --

Sent from an expensive device which will be obsolete in a few months! :D

Casey

On Jun 16, 2016 9:06 AM, mailto:amindfv@gmail.com> wrote:

foo :: Applicative f => f String foo = do x <- pure "this works" pure x

... but replace "pure x" with "pure $ x" and it doesn't typecheck: a monad instance is required!

Tom _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org mailto:Haskell-Cafe@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe

_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe

Reply

Sign in to reply online Use email software

amindfv＠gmail.com

12:49 p.m.

Wow, that's pretty unexpected. Another repro case: p = pure bar :: Applicative f => f a bar = do x <- pure "ok" p x ... gives the same error. So ApplicativeDo statements *must* end with exactly the statement "pure"?

El 16 jun 2016, a las 12:39, Joe Quinn escribió:

I expect it's that the root of the expression on the last line is ($) and not pure. ApplicativeDo has restrictions on what you can do to avoid allowing blocks that have to translate in terms of join/(>>=), and likely one of them is that the last line can't be an arbitrary thing of the right type.

...
On 6/16/2016 12:30 PM, amindfv@gmail.com wrote: They're the same:

...
:t \x -> pure x \x -> pure x :: Applicative f => a -> f a :t \x -> pure $ x \x -> pure $ x :: Applicative f => a -> f a

Tom

El 16 jun 2016, a las 12:24, KC escribió:

...
Think of the types of

pure x

And

pure $ x

-- --

Sent from an expensive device which will be obsolete in a few months! :D

Casey

...
On Jun 16, 2016 9:06 AM, wrote: foo :: Applicative f => f String foo = do x <- pure "this works" pure x

... but replace "pure x" with "pure $ x" and it doesn't typecheck: a monad instance is required!

Tom _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe

_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe

_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe

Reply

Sign in to reply online Use email software

Tom Ellis

12:56 p.m.

On Thu, Jun 16, 2016 at 12:49:03PM -0400, amindfv@gmail.com wrote:

Wow, that's pretty unexpected. Another repro case:

p = pure

bar :: Applicative f => f a bar = do x <- pure "ok" p x

... gives the same error. So ApplicativeDo statements *must* end with exactly the statement "pure"?

That's pretty strange since it has a perfectly good Applicative interpretation: fmap p (pure "ok") Tom

Reply

Sign in to reply online Use email software

Alexander V Vershilov

12:49 p.m.

Hello, Tom. I think you can find this ticket relevant: https://ghc.haskell.org/trac/ghc/ticket/11835 -- Alexander On 16 June 2016 at 19:08, wrote:

foo :: Applicative f => f String foo = do x <- pure "this works" pure x

... but replace "pure x" with "pure $ x" and it doesn't typecheck: a monad instance is required!

Tom _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe

-- Alexander

Reply

Sign in to reply online Use email software

3260

Age (days ago)

3260

Last active (days ago)

Download

6 comments

5 participants

tags

participants (5)

Alexander V Vershilov
amindfv＠gmail.com
Joe Quinn
KC
Tom Ellis