July 2010 - Haskell-Cafe

data type declaration
by Patrick Browne 25 Jul '10

25 Jul '10

Hi, I am trying to understand the data type declaration below. What is the relation between class C3 and the data type Address below? Where is such a technique used? Thanks, Pat module A where data Person = Person String Integer deriving Show data Employee = Employee String Integer deriving Show class C1 c1 where age :: c1 -> Integer instance C1 Person where age(Person "P1" 1) = 1 instance C1 Employee where age(Employee "E1" 1) = 1 class C1 c2 => C2 c2 where name :: c2 -> String instance C2 Person where name(Person "P2" 1) = "P2" instance C2 Employee where name(Employee "E2" 1) = "E2" class C2 c3 => C3 c3 a where address :: c3 -> a instance C3 Person String where -- ** What are the semantics or purpose of this construct. -- ** Is the type declared in the context of a class. data C3 c3 a => Address c3 a = Address c3 a instance C3 Person (Address String String) where This message has been scanned for content and viruses by the DIT Information Services E-Mail Scanning Service, and is believed to be clean. http://www.dit.ie

4 6

default function definitions
by Patrick Browne 25 Jul '10

25 Jul '10

Hi, I am studying the Haskell type class system as part of a language comparison thesis. At this point I am looking at how default function definitions are handled in classes. Functions are usually defined in instances, not classes. I appreciate that the code below may not be an appropriate way to write Haskell programs, but it does help me understand how defaults work. I am not trying to construct a running program. Any feedback on the questions below and defaults work in general would be appreciated. Pat module A where data Person = Person String Integer deriving Show data Employee = Employee String Integer deriving Show class C1 c1 where age :: c1 -> Integer -- add default impl, can this be defined only once at class level? -- Can this function be redefined in a *class* lower down the heirarchy? age(c1) = 1 -- Is it true that instances must exists before we can run function or make subclasses? instance C1 Person where instance C1 Employee where -- Is it true that C2 can inherit age, provided an instance of C1 exists class C1 c2 => C2 c2 where name :: c2 -> String name(c2) = "C2" instance C2 Person where instance C2 Employee where -- Is it true that C3 cannot override C1 or C2 existing defaults? -- Is it true that this must be done at instance level? -- class Cx c3 => C3 c3 where -- age(c3) = 3 -- This message has been scanned for content and viruses by the DIT Information Services E-Mail Scanning Service, and is believed to be clean. http://www.dit.ie

5 4

Type problems
by michael rice 25 Jul '10

25 Jul '10

This works: Prelude System.Random> do { randomRIO (1,6) >>= (\x -> putStrLn $ "Value = " ++ show x) } Value = 5 So does this: Prelude System.Random> do { x <- randomRIO (1,6); putStrLn $ "Value = " ++ show x } Value = 2 But not this: 1 import Control.Monad 2 import System.Random 3 4 foo :: IO () 5 foo = do 6 x <- randomRIO (1,6) 7 putStrLn $ "Value = " ++ show x foo.hs:6:18: Ambiguous type variable `t' in the constraints: `Num t' arising from the literal `1' at foo.hs:6:18 `Random t' arising from a use of `randomRIO' at foo.hs:6:7-21 Probable fix: add a type signature that fixes these type variable(s) Failed, modules loaded: none. Prelude System.Random> Or this: 1 import Control.Monad 2 import System.Random 3 4 foo :: IO () 5 foo = randomRIO (1,6) >>= (\x -> putStrLn $ "Value = " ++ show x) foo.hs:5:17: Ambiguous type variable `t' in the constraints: `Num t' arising from the literal `1' at foo.hs:5:17 `Random t' arising from a use of `randomRIO' at foo.hs:5:6-20 Probable fix: add a type signature that fixes these type variable(s) Failed, modules loaded: none. How to fix? Michael

5 8

Porting ELF statifier to Haskell!
by C K Kashyap 25 Jul '10

25 Jul '10

Hi, At my work we ran into a situation where we started wishing there was a way to take a dynamically linked executable and create a statically linked bundle out of it. Little bit of googling got me to statifier - http://statifier.sourceforge.net/statifier/main.html. The project seems a little old and when I tried it out on my 32bit RHEL5 box, the statically linked file did not run. So I thought it would be a good exercise to try and use Haskell's Elf module (Data.Elf) and attempt to build a statifier. Just wanted to understand if anyone's tried this before or have any advice for me. -- Regards, Kashyap

3 4

recommendation for (best) sqlite3 bindings
by Gour 25 Jul '10

25 Jul '10

Hello! We are looking for recommendation which Haskell bindings for sqlite3 to use for destkop GUI app where we want, among other things to store *.png and/or *.jpg images. (Yeah, I know about the hint to store iamges in the filesystem and just store filepaths in the db, but for portability reasons so that user can easily carry/backup database, we want everything stored in one file.) By looking at Hackage, it seems there are 3 candidates: 1) sqlite - bindings by Galois (http://hackage.haskell.org/package/sqlite) - looks quite complete & low-level interface 2) direct-sqlite - it says "It is not as complete as bindings-sqlite3 (1), but is slightly higher-level...it supports strings encoded as UTF8, and BLOBs represented as ByteStrings." and 3) HDBC-sqlite3 - higher level but without support for BLOBs. Now, based on the above it looks that 2) is the best one - not high-level as 2), but adding BLOBs support which, iirc, is missing in HDBC. Otoh, having highr-level abstraction ala HDBC is nice, although at the moment we believe that we won't have need to go to PostgreSQL since it means that setup would be greatly complicated for the end-user, so we're staying focused on Sqlite3. I know there are also Takusen & HSQL, but based on my past experiences when watching those two projects, it seems they aren not supported as well as the above itemized ones. Any recommendation? May I add that, according to the recent "Merge hsql and HDBC -- there can only be one!" thread, I can only "+1" for having slight less & more complete database packages. Sincerely, Gour -- Gour | Hlapicina, Croatia | GPG key: F96FF5F6 ----------------------------------------------------------------

3 5

build and use ghc's rts without a full unregistered ghc port?
by Korcan Hussein 24 Jul '10

24 Jul '10

Hi, I was just wondering if this is possible, I would like to use a gcc port which cross compiles to the PPC architecture (Wii DevkitPPC to be specifically: http://wiibrew.org/wiki/DevkitPPC) for a platform that is not POSIX compatible I believe (at least not fully or maybe I'm wrong entirely). Ideally I would make an unregistered port of GHC but the problem is because GHC doesn't support true cross compiling yet I need to bootstrap on to the target machine, which isn't running a POSIX environment. Still I did try to attempt to build with both the host and target set to 'powerpc-unknown-linux' and build it with devkitPPC, I did use a custom build.mk which should pull in all the required dependencies, defines, etc for devkitPPC. In the end I didn't get very far with it, it looked like missing posix functions. I didn't want to spend to much time trying to make it work this way (maybe i'm doing it wrong completely). It's quite possible that I could make this work more easily if I used a linux port for the wii but I would like to avoid the need of using linux on the wii because I guess most home-brew users don't have such a setup. So I thought maybe it's possible with less effort to build GHC's RTS and then compile Haskell in C with the RTS as a library dependency. Is this possible? maybe I'm missing something or doing something stupid? Thanks in advance. _________________________________________________________________ http://clk.atdmt.com/UKM/go/197222280/direct/01/ We want to hear all your funny, exciting and crazy Hotmail stories. Tell us now

2 3

"New FGL" naming survey
by Ivan Lazar Miljenovic 24 Jul '10

24 Jul '10

Whether or not the new FGL that Thomas Bereknyei and I are working on should keep the name was a semi-hot issue when we first mentioned the fact that we were working on a new version about a month ago. As such, I've created a survey here to try and find out what the Haskell community overall thinks we should call it: https://spreadsheets.google.com/viewform?formkey=dGpzMmFnUWY3Uktodk5wdHlLQk… More info can be found on the actual survey page. -- Ivan Lazar Miljenovic Ivan.Miljenovic(a)gmail.com IvanMiljenovic.wordpress.com

2 2

Re: [Haskell-cafe] FGL problem: cannot acces data constructor NodeMap
by Ivan Lazar Miljenovic 24 Jul '10

24 Jul '10

Once again cc-ing -cafe; Immanuel, can you please do "reply to all" rather than just "reply" in GMail? :p Immanuel Normann <immanuel.normann(a)googlemail.com> writes: > 2010/7/24 Ivan Lazar Miljenovic <ivan.miljenovic(a)gmail.com> > >> CC-ing haskell-cafe again: >> >> Immanuel Normann <immanuel.normann(a)googlemail.com> writes: >> >> > 2010/7/24 Ivan Lazar Miljenovic <ivan.miljenovic(a)gmail.com> >> >> > Some of my lnodes in the graph have by construction unique labels and I >> want >> > to lookup the lnode from a graph given only the label of that lnode. >> >> Well, from my understanding NodeMap doesn't do what you want, since it >> assumes that _all_ of the node labels are unique > > > I wouldn't bother when NodeMap maps also non-unique labels to single nodes, > because I only lookup nodes with unique labels, e.g. assume my label are of > the form > data Label = Unique String | Multi Label I'm of the suspicion that using NodeMaps in this fashion will have problem when considering nodes with the same label (even if you don't explicitly use it for those nodes), as NodeMap seems to assume that node labels are unique. For example, mkNode assumes node labels are unique. >> Furthemore, the way of using it seems to be analogous of how one would use a >> State monad >> rather than for actual interactive use throughout. >> > > Could you elaborate this further, please. http://hackage.haskell.org/packages/archive/fgl/5.4.2.3/doc/html/Data-Graph… and http://hackage.haskell.org/packages/archive/fgl/5.4.2.3/doc/html/Data-Graph… If you just want a mapping from labels to nodes, you may be better off using your own custom Map (and maybe even defining a new graph type that has such a Map as part of it). The successor to FGL-5.x (whatever we'll end up calling it) will allow you to have custom node types rather than using Ints. However, this will not work for you in this situation. The node type is just an _index_ type, analogous to the actual hash value in a hash{map,array} (as opposed to the value which has the hashing function applied upon it). It's there to provide a unique identifier for a particular node. Arguably, an abstract type would be even better as an Int as it will prevent you from doing silly things like adding node indices together, but that's beside the point. If you wish to find a node with a specific label, try using something like gsel: http://hackage.haskell.org/packages/archive/fgl/5.4.2.3/doc/html/Data-Graph… -- Ivan Lazar Miljenovic Ivan.Miljenovic(a)gmail.com IvanMiljenovic.wordpress.com

2 1

Re: [Haskell-cafe] FGL problem: cannot acces data constructor NodeMap
by Ivan Lazar Miljenovic 24 Jul '10

24 Jul '10

CC-ing haskell-cafe again: Immanuel Normann <immanuel.normann(a)googlemail.com> writes: > 2010/7/24 Ivan Lazar Miljenovic <ivan.miljenovic(a)gmail.com> > >> Immanuel Normann <immanuel.normann(a)googlemail.com> writes: >> > Afterall, my purpose is to get access to the map in a NodeMap and finally >> to >> > apply lookup to it. But I don't know how to access the map from a NodeMap >> > (the map selector isn't accessible either). >> >> You're not meant to; the point of NodeMap is to serve as a wrapper >> around the normal graph types, not for you to use directly (exactly the >> same as why the constructor for Map isn't exported). >> >> May I ask, however, why you want to use Data.Graph.Inductive.NodeMap? >> > > Some of my lnodes in the graph have by construction unique labels and I want > to lookup the lnode from a graph given only the label of that lnode. Well, from my understanding NodeMap doesn't do what you want, since it assumes that _all_ of the node labels are unique. Furthemore, the way of using it seems to be analogous of how one would use a State monad rather than for actual interactive use throughout. -- Ivan Lazar Miljenovic Ivan.Miljenovic(a)gmail.com IvanMiljenovic.wordpress.com

1 0

FGL problem: cannot acces data constructor NodeMap
by Immanuel Normann 24 Jul '10

24 Jul '10

Hi, I have a problem with the data constructor NodeMap Data.Graph.Inductive.NodeMap of the graph library fgl-5.4.2.3 (also fgl-5.4.2.2): I cannot access the data constructor NodeMap, as the ghci session shows: Prelude> :m Data.Graph.Inductive.NodeMap Prelude Data.Graph.Inductive.NodeMap> :t NodeMap <interactive>:1:0: Not in scope: data constructor `NodeMap' However, when I load the source directly it works: Prelude> :l Data/Graph/Inductive/NodeMap.hs [1 of 3] Compiling Data.Graph.Inductive.Graph ( Data/Graph/Inductive/Graph.hs, interpreted ) [2 of 3] Compiling Data.Graph.Inductive.Internal.FiniteMap ( Data/Graph/Inductive/Internal/FiniteMap.hs, interpreted ) [3 of 3] Compiling Data.Graph.Inductive.NodeMap ( Data/Graph/Inductive/NodeMap.hs, interpreted ) Ok, modules loaded: Data.Graph.Inductive.Internal.FiniteMap, Data.Graph.Inductive.Graph, Data.Graph.Inductive.NodeMap. *Data.Graph.Inductive.NodeMap> :t NodeMap NodeMap :: (Ord a) => FiniteMap a Node -> Int -> NodeMap a Why is that so? Afterall, my purpose is to get access to the map in a NodeMap and finally to apply lookup to it. But I don't know how to access the map from a NodeMap (the map selector isn't accessible either). Immanuel

2 1