December 2016 - Haskell-Cafe

Re: [Haskell-cafe] help with type variable escaping scope/passing lens as function argument
by Li-yao Xia 31 Dec '16

31 Dec '16

Hi Cody, I forgot to keep Haskell-cafe in copy. > Are you saying that my forall applies to both a and b, but not only a? And > that a is what (Lens (Record rs) Chicago -> (...)) expands to? That sounds about right. Try this: dateBetween :: (forall f. Functor f => (Chicago -> f Chicago) -> Record rs -> f (Record rs)) -> Day -> Day -> Pipe (Record rs) (Record rs) IO m Or, a bit refactored: type Lens s a = forall f. Functor f => (a -> f a) -> (s -> f s) dateBetween :: Lens (Record rs) Chicago -> Day -> Day -> Pipe (Record rs) (Record rs) IO m > Where does f1 come from? Does GHC just come up with it because my type > variable doesn't satisfy it? When you write this dateBetween :: forall (rs :: [*]) (m :: *) (f :: * -> *). (Functor f) => ((Chicago -> f Chicago) -> Record rs -> f (Record rs)) -> Day -> Day -> Pipe (Record rs) (Record rs) IO m dateBetween target start end = ... The parameter target is a monomorphic function, with f instantiated at the call site of dateBetween. This means that dateBetween can not assume anything about f apart from the fact that it is an instance of Functor, i.e., f must be abstract. Then we have: let targetDate = rget target r but this is the type of rget: rget :: (forall f . Functor f => (a -> f a) -> Record rs -> f (Record rs)) -> Record rs -> a rget expects a polymorphic function, while target is monomorphic. The compiler tries to unify "f" from the type signature of dateBetween, and "f" from the type signature of rget. To avoid confusion, the compiler does renaming, so that different variables (identified by where they are bound) have different names, and that's where "f1" comes from: rget :: forall a1 rs1. (forall f1 . Functor f1 => (a1 -> f1 a1) -> Record rs1 -> f1 (Record rs1)) -> Record rs1 -> a1 Applying rget to (target :: (Chicago -> f Chicago) -> Record rs -> f (Record rs)), the compiler tries to unify f and f1, and fails because they are "rigid", meaning, they can't be instantiated. f is rigid because it is abstract in the body of dateBetween, f1 is rigid because it is the type parameter of the polymorphic function rget expects. Li-yao On 12/31/2016 09:07 AM, Cody Goodman wrote: > Hi Li-yao, > > Are you saying that my forall applies to both a and b, but not only a? And > that a is what (Lens (Record rs) Chicago -> (...)) expands to? > > In either case, I don't understand what action I can take based on what was > you said and what I understand from it :/ > > Where does f1 come from? Does GHC just come up with it because my type > variable doesn't satisfy it? > > Thanks, > > Cody > > On Sat, Dec 31, 2016 at 2:54 AM, Li-yao Xia <li-yao.xia(a)ens.fr> wrote: > >> Hi Cody, >> >> You wrote something of the shape >> >> forall f. Functor f => a -> b >> >> which means >> >> forall f. Functor f => (a -> b) >> >> but not >> >> (forall f. Functor f => a) -> b >> >> which would be what (Lens (Record rs) Chicago -> (...)) expands to. >> >> Regards, >> Li-yao >> >> >> On 12/31/2016 08:45 AM, Cody Goodman wrote: >> >>> Thanks for the help Li-yao, but I still have a type error after using Rank >>> N Types. I have updated my code on github >>> https://github.com/codygman/frames-credit-card-trans-demo/bl >>> ob/master/src/Main.hs#L45() >>> and pasted below: >>> >>> dateBetween :: forall (rs :: [*]) (m :: *) (f :: * -> *). >>> (Functor f) => >>> ((Chicago -> f Chicago) -> Record rs -> f (Record rs)) >>> -> Day >>> -> Day >>> -> Pipe (Record rs) (Record rs) IO m >>> dateBetween target start end = P.filter go >>> where go :: Record rs -> _ >>> go r = let targetDate = (rget target r) :: Chicago >>> targetDate' = chicagoToZoned targetDate :: ZonedTime >>> targetDay = localDay (zonedTimeToLocalTime targetDate') >>> :: Day >>> in >>> targetDay >= start && targetDay < end >>> -- type error >>> -- src/Main.hs:53:39: error: … >>> -- • Couldn't match type ‘f’ with ‘f1’ >>> -- ‘f’ is a rigid type variable bound by >>> -- the type signature for: >>> -- dateBetween :: forall (rs :: [*]) m (f :: * -> *). >>> -- Functor f => >>> -- ((Chicago -> f Chicago) -> Record rs -> f >>> (Record rs)) >>> -- -> Day -> Day -> Pipe (Record rs) (Record rs) >>> IO m >>> -- at >>> /home/cody/source/frames-credit-card-trans-demo/src/Main.hs:45:45 >>> -- ‘f1’ is a rigid type variable bound by >>> -- a type expected by the context: >>> -- forall (f1 :: * -> *). >>> -- Functor f1 => >>> -- (Chicago -> f1 Chicago) -> Record rs -> f1 (Record rs) >>> -- at >>> /home/cody/source/frames-credit-card-trans-demo/src/Main.hs:53:34 >>> -- Expected type: (Chicago -> f1 Chicago) >>> -- -> Record rs -> f1 (Record rs) >>> -- Actual type: (Chicago -> f Chicago) -> Record rs -> f (Record >>> rs) >>> -- • In the first argument of ‘rget’, namely ‘target’ >>> -- In the expression: (rget target r) :: Chicago >>> -- In an equation for ‘targetDate’: >>> -- targetDate = (rget target r) :: Chicago >>> -- • Relevant bindings include >>> -- target :: (Chicago -> f Chicago) -> Record rs -> f (Record rs) >>> -- (bound at >>> /home/cody/source/frames-credit-card-trans-demo/src/Main.hs:51:13) >>> -- dateBetween :: ((Chicago -> f Chicago) >>> -- -> Record rs -> f (Record rs)) >>> -- -> Day -> Day -> Pipe (Record rs) (Record rs) >>> IO m >>> -- (bound at >>> /home/cody/source/frames-credit-card-trans-demo/src/Main.hs:51:1) >>> -- Compilation failed. >>> >>> On Sat, Dec 31, 2016 at 2:21 AM, Li-yao Xia <li-yao.xia(a)ens.fr> wrote: >>> >>> Hello Cody, >>>> >>>> Complete the type annotation, using RankNTypes >>>> >>>> dateBetween :: Lens Record Chicago -> ... >>>> >>>> GHC does not infer higher-rank types, because it is an undecidable >>>> problem >>>> in general, so you must annotate polymorphic parameters explicitly. >>>> >>>> Regards, >>>> Li-yao >>>> >>>> >>>> On 12/31/2016 07:56 AM, Cody Goodman wrote: >>>> >>>> I have the following code and error(also seen at >>>>> https://github.com/codygman/frames-credit-card-trans-demo/bl >>>>> ob/master/src/Main.hs#L28 >>>>> ): >>>>> >>>>> >>>>> -- between meaning on or after start but before end >>>>> dateBetween :: _ >>>>> -> Day >>>>> -> Day >>>>> -> Pipe (Record rs) (Record rs) IO r >>>>> dateBetween target start end = P.filter go >>>>> where go :: Record rs -> _ >>>>> go r = let targetDate = rget target r :: Chicago >>>>> targetDate' = chicagoToZoned targetDate :: ZonedTime >>>>> targetDay = localDay (zonedTimeToLocalTime >>>>> targetDate') >>>>> :: Day >>>>> in >>>>> targetDay >= start && targetDay < end >>>>> -- type error >>>>> -- src/Main.hs:48:38: error: … >>>>> -- • Couldn't match expected type ‘(Chicago -> f Chicago) >>>>> -- -> Record rs1 -> f (Record rs1)’ >>>>> -- with actual type ‘t’ >>>>> -- because type variable ‘f’ would escape its scope >>>>> -- This (rigid, skolem) type variable is bound by >>>>> -- a type expected by the context: >>>>> -- Functor f => (Chicago -> f Chicago) -> Record rs1 -> f >>>>> (Record >>>>> rs1) >>>>> -- at >>>>> /home/cody/source/frames-credit-card-trans-demo/src/Main.hs:48:33-45 >>>>> -- • In the first argument of ‘rget’, namely ‘target’ >>>>> -- In the expression: rget target r :: Chicago >>>>> -- In an equation for ‘targetDate’: >>>>> -- targetDate = rget target r :: Chicago >>>>> -- • Relevant bindings include >>>>> -- r :: Record rs1 >>>>> -- (bound at >>>>> /home/cody/source/frames-credit-card-trans-demo/src/Main.hs:48:12) >>>>> -- go :: Record rs1 -> Bool >>>>> -- (bound at >>>>> /home/cody/source/frames-credit-card-trans-demo/src/Main.hs:48:9) >>>>> -- target :: t >>>>> -- (bound at >>>>> /home/cody/source/frames-credit-card-trans-demo/src/Main.hs:46:13) >>>>> -- dateBetween :: t -> Day -> Day -> Pipe (Record rs) (Record >>>>> rs) >>>>> IO r >>>>> -- (bound at >>>>> /home/cody/source/frames-credit-card-trans-demo/src/Main.hs:46:1) >>>>> -- Compilation failed. >>>>> >>>>> Perhaps it's late, but I'm fairly sure I've done something like this >>>>> before. >>>>> >>>>> Thanks, >>>>> >>>>> Cody >>>>> >>>>> >>>>> >>>>> _______________________________________________ >>>>> Haskell-Cafe mailing list >>>>> To (un)subscribe, modify options or view archives go to: >>>>> http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe >>>>> Only members subscribed via the mailman list are allowed to post. >>>>> >>>>> >>>>> >>> >

2 1

help with type variable escaping scope/passing lens as function argument
by Cody Goodman 31 Dec '16

31 Dec '16

I have the following code and error(also seen at https://github.com/codygman/frames-credit-card-trans-demo/blob/master/src/M… ): -- between meaning on or after start but before end dateBetween :: _ -> Day -> Day -> Pipe (Record rs) (Record rs) IO r dateBetween target start end = P.filter go where go :: Record rs -> _ go r = let targetDate = rget target r :: Chicago targetDate' = chicagoToZoned targetDate :: ZonedTime targetDay = localDay (zonedTimeToLocalTime targetDate') :: Day in targetDay >= start && targetDay < end -- type error -- src/Main.hs:48:38: error: … -- • Couldn't match expected type ‘(Chicago -> f Chicago) -- -> Record rs1 -> f (Record rs1)’ -- with actual type ‘t’ -- because type variable ‘f’ would escape its scope -- This (rigid, skolem) type variable is bound by -- a type expected by the context: -- Functor f => (Chicago -> f Chicago) -> Record rs1 -> f (Record rs1) -- at /home/cody/source/frames-credit-card-trans-demo/src/Main.hs:48:33-45 -- • In the first argument of ‘rget’, namely ‘target’ -- In the expression: rget target r :: Chicago -- In an equation for ‘targetDate’: -- targetDate = rget target r :: Chicago -- • Relevant bindings include -- r :: Record rs1 -- (bound at /home/cody/source/frames-credit-card-trans-demo/src/Main.hs:48:12) -- go :: Record rs1 -> Bool -- (bound at /home/cody/source/frames-credit-card-trans-demo/src/Main.hs:48:9) -- target :: t -- (bound at /home/cody/source/frames-credit-card-trans-demo/src/Main.hs:46:13) -- dateBetween :: t -> Day -> Day -> Pipe (Record rs) (Record rs) IO r -- (bound at /home/cody/source/frames-credit-card-trans-demo/src/Main.hs:46:1) -- Compilation failed. Perhaps it's late, but I'm fairly sure I've done something like this before. Thanks, Cody

1 0

Netwire woes: creating a wire with mkGen_
by Francesco Ariis 31 Dec '16

31 Dec '16

Hello -cafe, a few days ago I posted this [1] in beginners, I'll pester you too before giving up. tl;dr: I want to create a wire which outputs a stream of 'q's when the letter Q is pressed and a stream of 'a' when the letter A is pressed. What I wrote is not working and I suspect I am not using mkGen_ correctly. Any pointers appreciated! [1] https://mail.haskell.org/pipermail/beginners/2016-December/017276.html

2 4

ANN: rapid, rapid-term: Rapid prototyping for Haskell
by esz＠posteo.de 31 Dec '16

31 Dec '16

Hi everybody, I'd like to officially announce my rapid prototyping suite unimaginatively called [Rapid] along with a helper library called [rapid-term]. The traditional approach to rapid prototyping with Haskell is to have a process running in the background that listens for source code changes and then recompiles/reloads and restarts the application. I'll call this process the *watcher*. This approach has a few drawbacks: * The watcher is separate from the main GHCi instance used for development. This means that the only access to the application state you have is through the interfaces you have explicitly defined in your code. * Restarting the application entails losing all ephemeral state, all network connections, all computed in-memory data, etc. * The watcher typically doesn't understand individual application components, so it's impossible to restart them individually. Rapid uses Chris Done's [foreign-store] library to provide a rapid prototyping framework for GHCi that lets you reload just the relevant parts of your application while keeping everything else running untouched, including all state, database/network connections, etc. It runs directly in the GHCi instance used for development, so you can talk to Rapid as well as your application from the REPL as it runs; no need to define debugging interfaces. The goal is to shorten the development cycle as much as possible. Another example use case is the development of interactive GUI applications. You can keep the GUI window open and preserve all of its run-time state (text you have entered, list items you have created, etc.), while reloading your application logic. It can also be used by batch applications that don't run for extended periods of time, for example to keep expensive resources in memory. For example load and parse your giant data file once and just reuse the resulting value across reloads. There is a detailed [Rapid tutorial] in the documentation of the `Rapid` module. It also walks you through Emacs haskell-mode integration. Rapid-term is a helper library for the development of applications that require an actual terminal. It spawns a separate terminal and keeps it open for the duration of the GHCi session. The application can then access this terminal and reuse it across runs. There is a [rapid-term tutorial] in the documentation of the `Rapid.Term` module, which also explains how to use rapid-term for rapid prototyping of Vty applications (same story as for GUI applications, but for Vty: keep your UI running, preserve your state, reload your application logic). Feedback, bug reports and code contributions are highly welcome! Greets, ertes [foreign-store]: https://hackage.haskell.org/package/foreign-store [rapid tutorial]: https://hackage.haskell.org/package/rapid-0.1.3/docs/Rapid.html [rapid-term tutorial]: https://hackage.haskell.org/package/rapid-term-0.1.0/docs/Rapid-Term.html [rapid-term]: https://hackage.haskell.org/package/rapid-term [rapid]: https://hackage.haskell.org/package/rapid

1 0

Doing the course "Functional Systems in Haskell"
by Damian Nadales 30 Dec '16

30 Dec '16

Hi, I want to follow the material in this course: http://www.scs.stanford.edu/16wi-cs240h/ Is anybody interested in studying the course material and completing the exercises? Has anybody tried to complete earlier versions of the course? Cheers, Damian.

1 0

hackagebot future ?
by Simon Michael 30 Dec '16

30 Dec '16

Hello all, hackagebot is the bot I've been maintaining for a few years which announces new hackage uploads in the #haskell IRC channel. I've tried to keep it "a well-behaved, non-annoying bot". It's configured to announce up to 5 uploads at a time, with at least a 5 minute interval between announcements. As our community grows, upload activity increases, and more megaprojects with many uploads appear, I've seen more complaints about the noise, and noticed channel ops muting it temporarily. A few weeks ago this happened, and also a possible problem with the correctness of the announcements or the feed was reported. I didn't have time to investigate, so I left it offline, inviting feedback if anyone missed it. Since then I've heard nothing. So, even though I personally like seeing the uploads, and the conversations and connections that sometimes result, I'm inclined to let hackagebot rest indefinitely. Or, would a new maintainer like to take it on ? If you have feelings about it one way or the other, please reply on-list. Thanks! -Simon

7 9

many haskell's mails are detected as spam on gmail
by Takenobu Tani 28 Dec '16

28 Dec '16

Hi, I'm using gmail. Recently, many haskell's mails are detected as spam on gmail. (ghc-devs, haskell-cafe, ghc-commit, ...) Does anyone know why? Do you know the workaround? Regards, Takenobu

13 17

References on GADT's
by Damian Nadales 28 Dec '16

28 Dec '16

Hi, I'm climbing up the standardized ladder of functional programming: https://news.ycombinator.com/item?id=13067741 The next topic on my list is to gain some understanding and practice on GADT's (competent section of the ladder). Do you have any resources that you would recommend? Are there any books that cover the topic? Thanks in advance, Damian.

1 0

Multi groupby with foldl' and Map.insertWithKey
by Olaf Klinke 28 Dec '16

28 Dec '16

I'd say the cleanest way to go is to split your Info data structure into an intermediate key and a value part like so: type Key = String data Value = Value {count :: !Int, healthTopics :: ![String]} keyvalue :: Info -> (Key,Value) keyvalue (Info _ h k) = (k,Value 1 [h]) and give Value a Monoid instance: instance Monoid Value where mempty = Value 0 [] mappend (Value c s) (Value c' s') = Value (c+c') (s++s') (Or you could use (Sum Int) instead of Int and have type Value = (Sum Int,[String]). Then the Monoid instance is derived for you.) Now you can transform a list of Infos into a Map using a generic function: makeMap :: [Info] -> M.Map Key Value makeMap = M.fromListWith mappend . map keyvalue Semantically, it's pretty identical to ALeX Kazik's suggestion of using foldl' and alter. Internally, fromListWith uses a strict fold, so strictness should be the same as using foldl'. Olaf

2 3

[NL-FP 2017] 2nd CFP: Dutch Functional Programming Day 2017
by J. Stutterheim 28 Dec '16

28 Dec '16

Second Call for Participation: Dear Friends of Functional Programming, There is still room for more participants for the FP day. Please register until December 31st so we can make arrangements for lunch and dinner. The next Dutch FP Day will be held on Friday, January 6 2017 at the Radboud University in Nijmegen. You are invited to some or all of: participate, give a presentation, join the dinner. For further information, please see the website. https://wiki.clean.cs.ru.nl/NL-FP_dag_2017 Feel free to forward this invitation to colleagues and students who might be interested. Best Regards on behalf of Markus Klinik

1 0