September 2015 - Haskell-Cafe

Parsing, infixes and on-the-fly precedence
by Jeffrey Brown 29 Sep '15

29 Sep '15

I imagine a parser that operates on (String,[String]) pairs, for which the second element allows the user to define precedence on the fly. For instance, it would read the pair (x,y) = ("a , b : c", [":" , ","]) as "(a , (b : c))" because y informs the parser that in x the colon binds first, then the comma. I know it is possible, because Haskell does it already! In Haskell we can define novel infix operators on the fly, including their precedence relative to each other. I don't know how, though, so that's my question. -- Jeffrey Benjamin Brown

4 5

Question, re: Typeclassopedia Ex. 4.2.1
by David Banas 29 Sep '15

29 Sep '15

Hi all, In trying to solve exercise 4.2.1 in Typeclassopedia, I’ve come up with the following: To prove: pure f <*> x = pure (flip ($)) <*> x <*> pure f Proof: pure (flip ($)) <*> x <*> pure f = (interchange) pure (flip ($)) <*> pure ($ f) <*> x = (homomorphism) pure (flip ($) ($ f)) <*> x = (definition of flip) pure ($ ($ f)) <*> x = (interchange) x <*> pure ($ f) = (interchange) (Is this step valid?) pure f <*> x Is the last step in my proof valid? Thanks, -db

2 1

Using Parsec with a recursive data as the stream
by Matteo Ferrando 28 Sep '15

28 Sep '15

Hello, I've posted this question in StackOverflow[1], but I thought this would be a good place to ask too: I'm writing an interpreter for functional programming language with with *mixfix operators*, just like Agda[2]. I used their paper[3] as reference. if_then_else_ : Bool -> a -> a -> a if True then x else _ = x if False then _ else x = x _/\_ : Bool -> Bool -> Bool True /\ True = True _ /\ _ = False So that means I had to run a parser (I used Alex/Happy), to get an AST, with this specific part (smaller than actual `Expr`): data Expr = Id String | Apply [Expr] | Forall Type Expr data Type = TypeBind String Expr And with this `Expr`, I have to run a second parser (which I intend to use Parsec) to do the following kind of processing: λ let example = Apply [Id "if", Id "a", Id "/\\", Id "b", Id "then", Id "c", Id "else", Id "d"] λ parseMixfix example Right (Apply [Id "if_then_else_",Apply [Id "_/\\_",Id "a",Id "b"],Id "c",Id "d"]) I started with a Parser that received a `Stream` of `[Expr]`, but this only accepts the lists in a `Apply`, and doesn't go deep in the *tree*, just parses on the top level. So I'm considering the option of instead of using `[Expr]` as the `Stream`, to use `Expr`, having to do the `Stream` instance for it; this is where I'm at: data Tok a = This a | Over (Tok a) deriving (Show) instance (Monad m) => Stream Expr m (Tok Expr) where uncons ex = check ex where check :: Monad m => Expr -> m (Maybe (Tok Expr, Expr)) check ex = case ex of Id s -> return $ Just (This (Id s), Apply []) Apply (x:xs) -> do mst <- check x return $ fmap (\(a, b) -> (Over a, b)) mst Which is using `data Tok` as kind of a Zipper breadcrumb (or at least I see it that way), to indicate how deep in the tree it comes from. I know this is not the correct code, but is for you folks to get the idea. I'm wondering if I'm on the right track or if there's a better solution for this problem. I'm also missing the `Forall` case here; that's because I was making tests with an `Id | Apply` only tree before. [1]: http://stackoverflow.com/posts/32831287 [2]: http://wiki.portal.chalmers.se/agda/pmwiki.php?n=ReferenceManual.Mixfix [3]: http://www.cse.chalmers.se/~nad/publications/danielsson-norell-mixfix.pdf

1 0

Fmap and Map for lists
by Strikingwolf2012 . 27 Sep '15

27 Sep '15

Does anyone have any idea why fmap (Functor map) and the list map are separate functions? I think it would make plenty sense for the Functor map to just be named map and have lists respond to that

2 1

Re: [Haskell-cafe] Parsing Atom feed link
by Francesco Ariis 27 Sep '15

27 Sep '15

On Sat, Sep 26, 2015 at 10:59:12AM +0300, fr33domlover wrote: > Hello, > > I'm using the package 'feed' to parse news feeds and I noticed it fails to > parse the item links of many feeds! I investigated and apparently here is why. > Many Atom feeds in practice seem to publish their item links like this: > > <link href="http://....."/> Checking the code [1] it looks like it is indeed the case linkRel = Right `fmap` pAttr "rel" e [1] http://hackage.haskell.org/package/feed-0.3.10.1/docs/src/Text-Atom-Feed-Im… > I looked at the examples at https://tools.ietf.org/html/rfc4287 and I just > checked Octopress (Ruby) and Ikiwiki (Perl) generated feeds. The RFC's examples > have item links without "alernate" (as the only item link provided) and both > generators I mentioned create feed item links without "alternate". > > > Should 'feed' be fixed to recognize them? Since you have done the homework and this is rfc compliant, have you tried sending an email to the maintainer [1]? It seems an uncontroversial change and they are active (last version released in 2015), I bet they are be willing to fix this. [1] http://hackage.haskell.org/package/feed

2 1

Solutions to Typeclassopedia Ex. 3.3.1?
by David Banas 27 Sep '15

27 Sep '15

Would anyone care to share their solution to exercise 3.3.1 in Typeclassopedia? Thanks, -db

4 4

Constrain Symbol to contain only alphabetic characters
by Alexander Vieth 27 Sep '15

27 Sep '15

Hi Café, Is it possible to do this? I looked through the source for GHC.TypeLits on hackage but found no leads. Thanks for any help, Alex

2 1

STM implementation
by Nicola Gigante 26 Sep '15

26 Sep '15

Hi all, I’m considering the idea of hacking into the STM implementation, but the source in rts/STM.c seems rather dense. Is there some documentation about the internals or some place to start to understand what’s going on? Thank you, Nicola

3 4

Bologna Haskell Meeting, 2015-10-17 @ 9:30
by m 26 Sep '15

26 Sep '15

Dear all, a Bologna Haskell Meeting will take place on Saturday, October 17 at 9:30. For details see here: https://www.eventbrite.it/e/biglietti-incontro-autunnale-haskell-ita-a-bolo… The page is in Italian and English: the main event is in Italian but many people speak English and we will try to include as many participants as possible. If you would like to give a talk, drop a note to Haskell_ITA(a)googlegroups.com (there is probably still a free slot at the beginning or at the end of the evening). In the same building there will also be an Elixir meeting: https://www.eventbrite.it/e/biglietti-bologna-elixir-erlang-meetup-18440536… (but we have a bigger and better meeting room ...) and participants to a meeting can also lurk the other event. Everybody is welcome! And if you are interested in participating to the event, please buy a ticket on Eventbrite (it is free) just for the meeting that you plan to attend to. Thanks, and see you in Bologna, if you happen to be around. P.S.: if anybody is travelling and needs further information about hotels, travel, etcetera, feel free to contact me directly, I'll be glad to help. -- .*. finelli /V\ (/ \) -------------------------------------------------------------- ( ) Linux: Friends dont let friends use Piccolosoffice ^^-^^ -------------------------------------------------------------- A mother takes twenty years to make a man of her boy, and another woman makes a fool of him in twenty minutes. -- Frost

1 0

Parsing Atom feed link
by fr33domlover 26 Sep '15

26 Sep '15

Hello, I'm using the package 'feed' to parse news feeds and I noticed it fails to parse the item links of many feeds! I investigated and apparently here is why. Many Atom feeds in practice seem to publish their item links like this: <link href="http://....."/> But in the 'feed' package, a link is taken to be an item link only if its relation is alternate, i.e. add an attribute rel="alternate". So for these feeds it returns 'Nothing' for 'getItemLink'. A quick workaround is to go over the links manually of course, which solves the issue locally, but I think it should be solved more generally. Is the problem in 'feed' which recognizes links incorrectly, or do those Atom feeds simply use the wrong way to publish item links? Are they required to use "alternate"? I looked at the examples at https://tools.ietf.org/html/rfc4287 and I just checked Octopress (Ruby) and Ikiwiki (Perl) generated feeds. The RFC's examples have item links without "alernate" (as the only item link provided) and both generators I mentioned create feed item links without "alternate". Should 'feed' be fixed to recognize them? From the RFC: atom:link elements MAY have a "rel" attribute that indicates the link relation type. If the "rel" attribute is not present, the link element MUST be interpreted as if the link relation type is "alternate". --fr33

1 0