June 2008 - Libraries

Galois move complete, Hackage back online
by Don Stewart 28 Jun '08

28 Jun '08

Galois has completed the move of its data center, and the services: hackage.haskell.org darcs.haskell.org are now back online. Thanks for your patience, everyone! Enjoy the new bandwidth. -- Don

1 0

Generic tries (long)
by Jamie Brandon 27 Jun '08

27 Jun '08

Hi, I'm writing a library for generic tries for the Summer of Code. The main point of this post is to get some feedback on the api but I'll briefly explain the idea. The point of a trie is to exploit the recursive nature of ADTs to save on expensive key comparisons and reduce space consumption. Hinze' original formulation is very elegant but results in very deep structures and is fairly inefficient. The normal trie optimisations (concatenating singleton maps, mainly) cant be applied to the generic version. I intend instead to encode ADTs as lists. This encoding can range from a simple walk of the ADT to creating a compressed, bit packing representation. The resulting tradeoff between encoding time and space usage should make the design fairly flexible. This will look something like: class Serial k where -- | Flattened form of key consists of a list of Nodes -- Node will be Int or UArray Int for compressed implementations type Node k -- | Flatten to a list of Nodes serialise :: k -> [Node k] -- | Reconstruct from a list of Nodes unserialise :: [Node k] -> k The api below is mostly cribbed from Adrian Hey's initial design. Guarantees about ordering will probably vary between maps. All ascending have descending version too. Strict versions of functions will be written where appropriate, I've omitted them here for brevity. Key reconstruction is likely to be expensive so it may make more sense to seperate foldrKeys and friends into a seperate class. Adrian has written instance of GMap for lists, UInts and Ord types so I can declare instance (Serial k) => GMap (ListGMap (Node k)) k where ... etc I should have a spot on code.haskell.org soon, at which point I'll put up a Haddock page with the most up to date version of the api. class GMap map where type k -- | The empty map. empty :: map a -- | Create a map with a single association. singleton :: k -> a -> map a -- | Create a map from a list of associations which /must/ be in ascending order of keys -- (with /no/ duplicate keys). If in doubt use one of the safer (but slower) 'fromAssocs' functions. fromAssocsAscending :: [(k,a)] -> map a -- | Return 'True' if the map contains no associations. isEmpty :: map a -> Bool -- | Return 'True' if the map contains exactly one association. isSingleton :: map a -> Bool -- | Return the value associated with the supplied key (if any). lookup :: k -> map a -> Maybe a -- | Insert a new association in the map if there is currently no value associated with the key. -- If there is a value associated with the key then replace it with the result of -- applying the supplied function to that value. insert :: (a -> a) -> k -> a -> map a -> map a -- | Delete the association for the supplied key (if any). delete :: k -> map a -> map a -- | This is a combined insert\/modify\/delete operation. The argument to the supplied function -- is ('Just' a) if there is a value (a) associated with the supplied key, otherwise 'Nothing'. -- If the return value is ('Just' a'), a' becomes the new value associated with the supplied key. -- If the return value is 'Nothing', the association for the supplied key (if any) is deleted. alter :: (Maybe a -> Maybe a) -> k -> map a -> map a -- | Evaluate the union of two maps. If the maps contain common keys then combine the -- values associated with those keys using the supplied function. The value arguments -- to this function are supplied in the same order as the map arguments. union :: (a -> a -> a) -> map a -> map a -> map a -- | Evaluate the intersection of two maps, combining common associations using the supplied function. intersection :: (a -> b -> c) -> map a -> map b -> map c -- | Evaluate the difference between two maps. For any key occuring in the second map, -- the corresponding association (if any) is deleted from the first map. -- The associated values in the second map are irrelevant. difference :: map a -> map b -> map a -- | Returns true if the keys in the first map are a subset of the keys in the second map. -- (This includes the case where the key sets are identical). Note that this function does -- not examine the associated values (which are irrelevant). See 'isSubmapOf' if you -- do want associated values examined. isSubsetOf :: map a -> map b -> Bool -- | Returns true if the keys in the first map are a subset of the keys in the second map -- and the corresponding function always returns true when applied to the values associated -- with matching keys. isSubmapOf :: (a -> b -> Bool) -> map a -> map b -> Bool -- | Apply the supplied function to every associated value in the map. map :: (a -> b) -> map a -> map b -- | Apply the supplied function to every association in the map, and use the result -- as the new associated value for the corresponding key. mapWithKey :: (k -> a -> b) -> map a -> map b -- | Delete associations for which the supplied predicate returns 'False' when applied to -- the associated value. filter :: (a -> Bool) -> map a -> map a -- | Fold right over the list of elements in ascending order of keys. -- See 'foldrElemsAscending'' for a strict version of this function. foldrElemsAscending :: (a -> b -> b) -> map a -> b -> b -- | Fold right over the list of keys in ascending order. -- See 'foldrKeysAscending'' for a strict version of this function. foldrKeysAscending :: (k -> b -> b) -> map a -> b -> b -- | Fold right over the list of associations in ascending order of keys. -- See 'foldrAssocsAscending'' for a strict version of this function. foldrAssocsAscending :: (k -> a -> b -> b) -> map a -> b -> b -- | Fold over elements in un-specified order using /unboxed/ Int accumulator (with GHC). -- Defaults to boxed Int for other Haskells. Typically used for counting functions. -- Implementations are free to traverse the map in any order. -- The folded function is always applied strictly. foldElemsUINT :: (a -> UINT -> UINT) -> map a -> UINT -> UINT In addition there a few functions which are useful for groups of maps or nested maps. -- | Add the number of associations in a map to the supplied /unboxed/ Int (with GHC). -- Defaults to boxed Int for other Haskells. addSize :: map a -> UINT -> UINT -- | Find the value associated with the supplied key (if any) and return the result -- of applying the supplied continuation function to that value. Useful for nested lookup. lookupCont :: (a -> Maybe b) -> k -> map a -> Maybe b -- | Reject empty maps (return Nothing). nonEmpty :: map a -> Maybe (map a) The following functions are useful internally as most of the maps are defined in terms of simpler maps. Also see this thread on the need for unionMaybe. -- | Similar to 'insert', but the association is deleted if the supplied function returns 'Nothing'. -- (The supplied function is always applied strictly.) insertMaybe :: (a -> Maybe a) -> k -> a -> map a -> map a -- | Find the value associated with the supplied key (if any) and apply the supplied function -- to that value. Delete the association if the result is 'Nothing'. Replace the old value with -- the new value if the result is ('Just' something). -- (The supplied function is always applied strictly.) deleteMaybe :: (a -> Maybe a) -> k -> map a -> map a -- | Evaluate the union of two maps, but delete combined associations from the result map -- if the combining function returns 'Nothing'. -- (The combining function is always applied strictly.) unionMaybe :: (a -> a -> Maybe a) -> map a -> map a -> map a -- | Evaluate the intersection of two maps, but delete combined associations from the result map -- if the combining function returns 'Nothing'. -- (The combining function is always applied strictly.) intersectionMaybe :: (a -> b -> Maybe c) -> map a -> map b -> map c -- | Difference with a combining function. If the combining function returns -- @Just a@ then the corresponding association is not deleted from the result map -- (it is retained with @a@ as the associated value). differenceMaybe :: (a -> b -> Maybe a) -> map a -> map b -> map a -- | Apply the supplied function to every associated value in the map. -- If the result is 'Nothing' then the delete the corresponding association. -- (The supplied function is always applied strictly.) mapMaybe :: (a -> Maybe b) -> map a -> map b Thanks Jamie

9 27

Hackage and darcs.haskell.org downtime
by Don Stewart 26 Jun '08

26 Jun '08

Galois Inc is moving premises tonight, which means all our servers will be going offline in a few hours, physically moved several miles, and will reappear on a new high speed connection sometime in the following 24 hours. Expect up to 24 hours downtime for all systems hosted on 'monk', including: hackage.haskell.org darcs.haskell.org This will not affect: code.haskell.org www.haskell.org After the move, expect much higher bandwidth to darcs.haskell.org :) -- Don

1 0

Can't find cabal.exe on windows
by Aemon Cannon 26 Jun '08

26 Jun '08

I installed Cabal-1.4.01 with ghc 6.8.3 on Windows XP SP2 using the following commands: ghc --make Setup.hs Setup configure Setup build Setup install Everything seemed to work perfectly, but I can't seem to find cabal.exe The README seems to indicate that it'd be installed in HOME/.cabal, but this directory was not created. Any ideas? Thanks, Aemon Cannon

3 3

PROPOSAL: Make Applicative a superclass of Monad
by Ashley Yakeley 25 Jun '08

25 Jun '08

http://hackage.haskell.org/trac/ghc/ticket/2392 Make Applicative (in Control.Applicative) a superclass of Monad (in Control.Monad). Rename members of Applicative and other functions, to avoid unnecessary duplication. class Functor f => Applicative f where return :: a -> f a ap :: f (a -> b) -> f a -> f b liftA2 :: Applicative f => (a -> b -> c) -> f a -> f b -> f c liftA2 f a b = ap (fmap f a) b (>>) :: (Applicative f) => f a -> f b -> f b (>>) = liftA2 (const id) -- etc. class Applicative m => Monad m where (>>=) :: m a -> (a -> m b) -> m b fail :: String -> m a fail s = error s I do not have a patch for this change, as it involves mucking around with GHC.Base, and I strongly suspect also fiddling with the GHC source. I'd like to know if it's feasible first. Discussion period: a month? -- Ashley Yakeley

11 18

Package spam :-)
by Adrian Hey 24 Jun '08

24 Jun '08

Hello folks in general and GwernBranwen in particular :-) I don't want to seem ungratful, but I see a lot of the old "new" collections repository has been uploaded to Hackage as separate packages but it really shouldn't be there IMHO. AFAIK most of this stuff has never been properly tested and is in any case not being actively worked on or maintained any more. AvlTree and COrdering have been thoroughly tested and are already in Hackage as separate packages. These are not being worked on or maintained by anyone either, but they are the latest official releases that anyone interested should use. The generalised trie stuff is what Jamie Brandon is working on for his GSOC project. The relevant packages are: Data-Tree-AVL 0.4 Data-COrdering 0.4 Data-Trie-General 0.4 collections-base-instances 0.4 collections-rangedsets-instances 0.4 collections-api 0.4 enumsets 0.4 collections-avl 0.4 Data-Trie-General-OrdGT 0.4 collections 0.4 Regards -- Adrian Hey

10 18

Cabal 1.4.0.1 on Debian Stable/GHC 6.6
by Jochem Berndsen 24 Jun '08

24 Jun '08

Hi, On a Debian stable (etch) system, I have the following packages installed: jochem@huygens:/tmp/Cabal-1.4.0.1$ dpkg -l | grep ghc ii ghc6 6.6-3 GHC - the Glasgow Haskell Compilation system ii ghc6-prof 6.6-3 Profiling libraries for the Glasgow Haskell ii libghc6-filepath-dev 0.11-2 GHC 6 library for manipulating FilePath's ii libghc6-mtl-dev 1.0-3 Haskell monad transformer library for GHC ii libghc6-network-dev 2.0-3 Haskell network library for GHC jochem@huygens:/tmp/Cabal-1.4.0.1$ ghc-pkg list /usr/lib/ghc-6.6/package.conf: Cabal-1.1.6, (FilePath-0.11), base-2.0, filepath-1.1.0.0, (ghc-6.6), haskell98-1.0, mtl-1.0, network-2.0, parsec-2.0, readline-1.0, regex-base-0.71, regex-compat-0.71, regex-posix-0.71, rts-1.0, stm-2.0, template-haskell-2.0, unix-1.0 jochem@huygens:/tmp/Cabal-1.4.0.1$ I compiled Setup.hs, ran './Setup configure', './Setup build' and finally as root './Setup install'. Then ghc-pkg list gave: jochem@huygens:/tmp/bimap-0.2.2$ ghc-pkg list /usr/lib/ghc-6.6/package.conf: Cabal-1.1.6, Cabal-1.4.0.1, (FilePath-0.11), base-2.0, filepath-1.1.0.0, (ghc-6.6), haskell98-1.0, mtl-1.0, network-2.0, parsec-2.0, readline-1.0, regex-base-0.71, regex-compat-0.71, regex-posix-0.71, rts-1.0, stm-2.0, template-haskell-2.0, unix-1.0 jochem@huygens:/tmp/bimap-0.2.2$ But Cabal seems to be broken; for example, with the bimap package from Hackage: jochem@huygens:/tmp/bimap-0.2.2$ runghc Setup.lhs configure Configuring bimap-0.2.2... Setup.lhs: ghc version >=6.4 is required but the version of /usr/bin/ghc could not be determined. jochem@huygens:/tmp/bimap-0.2.2$ Cabal 1.2.3.0 seems to be working. What am I doing wrong? Or is this an unsupported combination? -- Jochem Berndsen | jochem(a)functor.nl GPG: 0xE6FABFAB

2 5

Re: [Haskell-cafe] ANNOUNCE: Pipe 1.0
by Isaac Dupree 23 Jun '08

23 Jun '08

Don Stewart wrote: > Interesting. > Does it depend on an unreleased version of the process library? by the way, is there a policy for when new versions of packages maintained by libraries(a)h.o are *released*? Or do patches just collect in the darcs repository until they're picked up by some major release like GHC's? I would tend to think that for any change nontrivial enough to be discussed on libraries@, why not quickly release the newer version with an appropriately incremented version number? -Isaac

2 1

[richardg@richardg.name: HUnit]
by Don Stewart 20 Jun '08

20 Jun '08

Forwarded from glasgow-haskell ----- Forwarded message from Richard Giraud <richardg(a)richardg.name> ----- Date: Wed, 18 Jun 2008 23:04:01 -0600 From: Richard Giraud <richardg(a)richardg.name> To: glasgow-haskell-users(a)haskell.org Subject: HUnit I've converted the literate Haskell documentation and filled in the gaps for HUnit. I've posted the darcs repository with the changes and the new documentation at http://richardg.name/ Please email me with any feedback and/or patches that you might have. In a bit, I'll submit the patches for inclusion into the GHC repository. Richard G. _______________________________________________ Glasgow-haskell-users mailing list Glasgow-haskell-users(a)haskell.org http://www.haskell.org/mailman/listinfo/glasgow-haskell-users ----- End forwarded message -----

2 1

Cabal-1.4: library-dirs and include-dirs missing from package description after install
by Alistair Bayley 20 Jun '08

20 Jun '08

I'm trying to build and install Takusen 0.8.2 with a new .cabal and Setup, using Cabal-1.4.0.0. The Setup program writes out an updated Takusen.buildinfo (below) from the postConf hook. However, the library-dirs and include-dirs fields are not included in the package description when installed. Any ideas as to what I'm doing wrong? Alistair --------------- Takusen.buildinfo --------------------- buildable: True build-tools: odbcconf -any, sqlplus -any, pg_config -any, sqlite3 -any cpp-options: cc-options: ld-options: --enable-stdcall-fixup pkgconfig-depends: frameworks: c-sources: extensions: CPP extra-libraries: odbc32 oci pq sqlite3 extra-lib-dirs: "C:\\Program Files\\sqlite3\\" C:/PROGRA~1/POSTGR~1/8.1/lib C:\oracle\product\10.2.0\client_1\bin includes: install-includes: include-dirs: "C:\\Program Files\\sqlite3\\" C:/PROGRA~1/POSTGR~1/8.1/include C:/PROGRA~1/POSTGR~1/8.1/include/server C:\oracle\product\10.2.0\client_1\oci\include hs-source-dirs: . other-modules: Database.InternalEnumerator ghc-prof-options: -prof -auto-all ghc-shared-options: -prof -auto-all ghc-options: hugs-options: nhc98-options: jhc-options: ------------------------------------------------------------------------ ghc-pkg describe Takusen-0.8.2: name: Takusen version: 0.8.2 license: BSD3 copyright: 2003-2008, Alistair Bayley, Oleg Kiselyov maintainer: alistair(a)abayley.org, oleg(a)pobox.com stability: experimental homepage: http://darcs.haskell.org/takusen package-url: http://darcs.haskell.org/takusen description: Takusen is a DBMS access library. ... <trimmed> category: Database author: Alistair Bayley, Oleg Kiselyov exposed: True exposed-modules: Database.ODBC.Enumerator Database.ODBC.OdbcFunctions Database.Oracle.Enumerator Database.Oracle.OCIConstants Database.Oracle.OCIFunctions Database.PostgreSQL.Enumerator Database.PostgreSQL.PGFunctions Database.Sqlite.Enumerator Database.Sqlite.SqliteFunctions Database.Enumerator Database.Util Database.Stub.Enumerator Control.Exception.MonadIO Foreign.C.UTF8 hidden-modules: Database.InternalEnumerator import-dirs: "C:\\Program Files\\Haskell\\Takusen-0.8.2\\ghc-6.8.1" library-dirs: "C:\\Program Files\\Haskell\\Takusen-0.8.2\\ghc-6.8.1" hs-libraries: HSTakusen-0.8.2 extra-libraries: odbc32 oci pq sqlite3 extra-ghci-libraries: include-dirs: includes: depends: base-3.0.0.0 mtl-1.1.0.0 old-time-1.0.0.0 time-1.1.2.0 hugs-options: cc-options: ld-options: --enable-stdcall-fixup framework-dirs: frameworks: haddock-interfaces: "C:\\Program Files\\Haskell\\doc\\Takusen-0.8.2\\html\\Takusen.haddock" haddock-html: "C:\\Program Files\\Haskell\\doc\\Takusen-0.8.2\\html"

2 5