February 2011 - Haskell-Cafe

Re: [Haskell-cafe] Froglingo – a functional programming language
by kevin＠froglingo.com 15 Feb '11

15 Feb '11

Upon a few requests, a brief view about how you would do with Froglingo is given as following: With Froglingo, users' main task is to organize business data and logic in the EP data model, i.e., the way that higher-order functions are related to each others. The remaining task is to specify constraints for data entries and to generate reports (queries). Developing data-intensive software applications in Froglingo is analogous to database management in a relational DBMS, where users only need to define data schemes, enter data, and specify data queries. The main difference is the expressive powers: the relational DBMS is for tables and the relational albegra, and the Froglingo is for total recursive functions and a class of total recursive functions. Note that a class of total recursive functions get rid of non-termination processes and keep the complete semantics for all the meaningful software applications. Depending on business needs, users can specify different report formats such as Froglingo expressions, HTML/XML, or any textual based formats. Froglingo has its own web server such that users can send requests through HTTP over internet. It manages files as data. It has built-in access control operators to protect data as if a file system has its own operators to protect files. The roles of programmers in Froglingo are consolidated with DBAs' because there is no longer "source code" separated from database (every data or logic is in database). Type-free is the most important difference of Froglingo from traditional programming languages in practice. In other words, users don't need to define their own types (e.g., data structures or classes) for software applications. The subtypes of applications are specified as constraints via Froglingo built-in operators as if employee data and salary data were specified in tables of a relational DBMS. Unlike other type-free systems such as the lambda-calculus, the EP data model guarantees bug-free (non-termination). Please let me know if you have any questions. Thank you. Kevin -----Original Message----- From: kevin(a)froglingo.com [mailto:kevin@froglingo.com] Sent: Tuesday, February 8, 2011 07:21 AM To: 'Haskell Cafe' Subject: [Haskell-cafe] Froglingo – a functional programming language Hi, I have recently finished a paper about Froglingo, a programming language designed to make it easier to write data-intensive business applications. It is a monolith consolidating programming, database/file management, data exchange, web server, and access control. What makes Froglingo unique is a sub language that is the first known as semantically equivalent to a class of TOTAL recursive functions. Froglingo shares with Haskell a declarative approach to programming, and strong mathematical foundations, so I thought that people on this list might find it of some interest. You can find the paper at: http://www.froglingo.com/FroglingoPL.pdf. If you have any thoughts about it, I’d love to hear from you. Thank you for your attention. Kevin

1 0

Headlines on haskell.org
by Michael Snoyman 15 Feb '11

15 Feb '11

Hi everyone, I'm not quite sure to whom to address this, but it is with regards to the "Headlines" section at the bottom of haskell.org. It has not been updated for 2011 yet, which I can't imagine looks very good to new users. Is anyone interested in/able to update it? If I may be so bold, I would like to offer up Warp as a possible headline. Are there any objections to this? Michael

5 4

rewrite rules to specialize function according to type class?
by Patrick Bahr 15 Feb '11

15 Feb '11

Hi all, I am trying to get a GHC rewrite rule that specialises a function according to the type of the argument of the function. Does anybody know whether it is possible to do that not with a concrete type but rather a type class? Consider the following example: > class A a where > toInt :: a -> Int > {-# NOINLINE toInt #-} > class B a where > toInt' :: a -> Int The idea is to use the method of type class A unless the type is also an instance of type class B. Let's say that Bool is an instance of both A and B: > instance A Bool where > toInt True = 1 > toInt False = 0 > instance B Bool where > toInt' True = 0 > toInt' False = 1 Now we add a rule that says that if the argument to "toInt" happens to be an instance of type class B as well, use the method "toInt'" instead: > {-# RULES > "toInt" forall (x :: B a => a) . toInt x = toInt' x > #-} Unfortunately, this does not work (neither with GHC 6.12 or GHC 7.0). Expression "toInt True" gets evaluated to "1". If the rewrite rule is written with a concrete type it works as expected: > {-# RULES > "toInt" forall (x :: Bool) . toInt x = toInt' x > #-} Now "toInt True" is evaluated to "0". Am I doing something wrong or is it not possible for GHC to dispatch a rule according to type class constraints? Thanks, Patrick

6 13

problems with HDBC
by Kurt Norwood 15 Feb '11

15 Feb '11

Hi, I am getting some very cryptic error messages while trying to compile some simple code using hdbc. I am able to run the following lines in ghci without any problems. I am using ghc 6.12.1 on Ubuntu 10.10. Thanks in advance, Kurt ################## # Contents of the file ################## import Database.HDBC.Sqlite3 import Database.HDBC main = dosomething dosomething :: IO () dosomething = do conn <- connectSqlite3 "testdb" run conn "create table test (id integer)" [] run conn "insert into test (id) values (6)" [] commit conn disconnect conn ############## # Error message ############## > ghc testsql.hs testsql.o: In function `s1eF_info': (.text+0x2e): undefined reference to `HDBCzmsqlite3zm2zi3zi1zi0_DatabaseziHDBCziSqlite3ziConnectionImpl_zdfIConnectionConnection_closure' testsql.o: In function `s1eG_info': (.text+0x6e): undefined reference to `HDBCzmsqlite3zm2zi3zi1zi0_DatabaseziHDBCziSqlite3ziConnectionImpl_zdfIConnectionConnection_closure' testsql.o: In function `s1eJ_info': (.text+0x16e): undefined reference to `HDBCzmsqlite3zm2zi3zi1zi0_DatabaseziHDBCziSqlite3ziConnectionImpl_zdfIConnectionConnection_closure' testsql.o: In function `s1eM_info': (.text+0x272): undefined reference to `HDBCzmsqlite3zm2zi3zi1zi0_DatabaseziHDBCziSqlite3ziConnectionImpl_zdfIConnectionConnection_closure' testsql.o: In function `s1ho_info': (.text+0x379): undefined reference to `HDBCzmsqlite3zm2zi3zi1zi0_DatabaseziHDBCziSqlite3ziConnection_connectSqlite3_closure' testsql.o: In function `s1ho_info': (.text+0x489): undefined reference to `__stginit_HDBCzm2zi2zi6zi1_DatabaseziHDBC_' testsql.o: In function `s1ho_info': (.text+0x493): undefined reference to `__stginit_HDBCzmsqlite3zm2zi3zi1zi0_DatabaseziHDBCziSqlite3_' testsql.o: In function `s1eF_info': (.text+0x36): undefined reference to `HDBCzm2zi2zi6zi1_DatabaseziHDBCziTypes_disconnect_info' testsql.o: In function `s1eG_info': (.text+0x76): undefined reference to `HDBCzm2zi2zi6zi1_DatabaseziHDBCziTypes_commit_info' testsql.o: In function `s1eJ_info': (.text+0x176): undefined reference to `HDBCzm2zi2zi6zi1_DatabaseziHDBCziTypes_run_info' testsql.o: In function `s1eM_info': (.text+0x27a): undefined reference to `HDBCzm2zi2zi6zi1_DatabaseziHDBCziTypes_run_info' testsql.o: In function `s1eN_srt': (.data+0x8): undefined reference to `HDBCzmsqlite3zm2zi3zi1zi0_DatabaseziHDBCziSqlite3ziConnectionImpl_zdfIConnectionConnection_closure' testsql.o: In function `s1ho_srt': (.data+0x18): undefined reference to `HDBCzmsqlite3zm2zi3zi1zi0_DatabaseziHDBCziSqlite3ziConnection_connectSqlite3_closure' collect2: ld returned 1 exit status

2 2

Fwd: ANN: vector-buffer package 0.1
by Alexander McPhail 15 Feb '11

15 Feb '11

Hi list, Could someone explain why the error pointed out by Luke occurred? From: Luke Palmer <lrpalmer(a)gmail.com> I think this would be a very good question for the list. Don't worry, they're nice helpful folks. On Mon, Feb 14, 2011 at 10:10 PM, Alexander McPhail <haskell.vivian.mcphail(a)gmail.com> wrote: > Hi Roman, > > > Can you explain why the following code doesn't work? 'unsafePerformIO' is > used all the time in hmatrix. For example, adding two vectors, we create a > new Vector then use that as the target for an FFI C call. > > Is the difference because I'm using mutable Storable Vectors? I would have > thought that unsafeFreeze (at the end) would make sure the problem reported > by Luke wouldn't occur. > > Is there a problem with laziness in the let binding of Luke's example? > > I note that in Data.Vector.Storable there is a pure 'convert' function, > which is essentially what I am trying to emulate. > > -- | convert to a vector > toVector :: Storable a => Buffer a -> (V.Vector a) > toVector (B o n v) = unsafePerformIO $ do > w <- M.new n > i <- readIORef o > M.unsafeWith v $ \p -> > M.unsafeWith w $ \q -> do > let n' = n-i > copyArray q (p `advancePtr` i) n' > if i /= 0 > then copyArray (q `advancePtr` n') p i > else return () > V.unsafeFreeze w > {-# INLINE toVector #-} > > > Vivian > > From: Luke Palmer <lrpalmer(a)gmail.com> > > This interface is an outlaw. > > main = do > buf <- newBuffer 10 :: IO (Buffer Int) > pushNextElement buf 1 > let v1 = V.toList (toVector buf) > v2 = V.toList (toVector buf) > print v1 > pushNextElement buf 2 > print v2 > > Despite v1 and v2 being defined to equal the exact same thing, this > program prints two distinct lines. > > toVector depends on the current state of the buffer. If this is to be > a law-abiding interface, toVector must return a value in IO: > > toVector :: (Storable a) => Buffer a -> IO (Vector a) > > Luke >

2 1

ANN: vector-buffer package 0.1
by Alexander McPhail 14 Feb '11

14 Feb '11

Hi, I have uploaded a small package, vector-buffer, to hackage. It provides a buffer that can be turned into a Data.Vector.Storable. The mapM* functions map from the oldest element, not the first. Similarly for the derived Vector. Feature requests etc. welcome. Vivian

2 2

lhs2tex and line numbers
by Mitar 14 Feb '11

14 Feb '11

Hi! How could I convince lhs2tex to add in poly mode line numbers before each code line in code block? Mitar

2 2

Re: [Haskell-cafe] Proving correctness
by kevin＠froglingo.com 14 Feb '11

14 Feb '11

On 14.02.2011, at 03:03 PM, Pavel Perikov [perikov(a)gmail.com] wrote: Sorely, Haskell can't prove logic with it. No predicates on values, guarantee that proof is not _|_. Haskell makes bug free software affordable, that's true. But it's not a proof assistant. Being equivalent to a class of total recursive functions, the EP (Enterprise-Participant) data model can assist users to enter arbitrary and meaningful expressions only. It is a new addition to programming languages. Check out here: http://www.froglingo.com/FroglingoPL.pdf Kevin

1 0

Problems with (projects|community).haskell.org
by Erik de Castro Lopo 14 Feb '11

14 Feb '11

Hi all, Still a couple of problems with these servers. Firstly, community.haskell.org shows the default Apache "It works" page. It would be nice to have something better there. Secondly the mailman web interface on projects.haskell.org [0] is giving a Service Temporarily Unavailable message (and has been for a couple of days). Cheers, Erik [0] http://projects.haskell.org/cgi-bin/mailman/admindb/haskell-llvm -- ---------------------------------------------------------------------- Erik de Castro Lopo http://www.mega-nerd.com/

4 3

ghci command line execution
by Thomas Davie 14 Feb '11

14 Feb '11

A while ago I remember someone showing me some tool, I *think* ghci that allowed you to pass it a function of type String -> String as an input, and have it simply run that function on stdin (presumably using interact) to achieve useful things like this... $ cat myFile.txt | ghci -e 'unlines . zipWith (\x y -> show x ++ " " ++ y) [1..] . lines' Does anyone know what it was that I saw being used here? Thanks Tom Davie

4 3