November 2001 - Glasgow-haskell-users

RE: profiling behavior
by Simon Marlow 13 Nov '01

13 Nov '01

> If module A imports B, and if B was compiled without the > profiling options > (but A was), is the final program expected to abort (illegal > operation or > something like that)? If it is, why? You must compile (and link) the whole program with -prof, because that option changes certain fundamental properties of the virtual machine. If you link modules compiled in different ways, the behaviour is unspecified ;-) Cheers, Simon

2 1

Input Processing Loops in Haskell
by Mark Conway Wirt 13 Nov '01

13 Nov '01

I apologize if this is a bit off-topic -- I'm asking it here because if what I want to do is possible, it may only be possible with ghc extensions, as opposed to vanilla Haskell...... In many types of applications, it's common to have a loop which loops infinitely, of at least until some condition is met. For example, in C-like syntax: for (;;;) { a = get_input(); if (a == GET_OUT) break; do_comthing(); } the above code could be used in an input-processing loop. The question is, is something like this possible in Haskell? I realize that you can use recursive calls to do something similar: f a | a == GET_OUT = a | otherwise = f ( g a) The problem is stack-growth. I can't figure out a way (I'm sort of a newbie) to so something like the above without growing my stack infinitely. I've seen something called "tail-recursion" discussed in comp.lang.functional that *may* be relevant, but I can't find any concise explanation of this concept. So, is this possible in either Haskell or Haskell+ghc extensions? TIA, Mark.

1 0

profiling behavior
by Andre W B Furtado 13 Nov '01

13 Nov '01

If module A imports B, and if B was compiled without the profiling options (but A was), is the final program expected to abort (illegal operation or something like that)? If it is, why? -- Andre

1 0

Broken link
by Andre W B Furtado 12 Nov '01

12 Nov '01

The link to "The Graph Visualisation Tool daVinci", in GHC profiling documentation (http://www.haskell.org/ghc/docs/latest/set/prof-xml-tool.html) seems to be broken. I think the link should point to http://www.informatik.uni-bremen.de/daVinci/ -- Andre

1 0

suggestion for: hslibs/hssource/
by Bernard James POPE 12 Nov '01

12 Nov '01

Hi, I have a small suggestion/request for the haskell parsing/printing utils in hslibs/hssource/ I think it would be nice to have SrcLocs for lambda bindings. Currently SrcLocs are available for all other occurrences of patterns, except for lambdas. The change required is quite small. Cheers, Bernie.

1 0

hdirect-0.18 comserv example
by Janna Khegai 11 Nov '01

11 Nov '01

I've tried to build the comserv example with hDirect-0.18 and ghc-5.02 under Windows ME. I've encountered the problems similar to ones mentioned earlier in: http://www.haskell.org/pipermail/glasgow-haskell-users/2001-October/000974.… : >problems generating the type library: it is simply not created. and http://www.haskell.org/pipermail/glasgow-haskell-users/2001-October/000979.… : > a problem linking the DLL, I always get this error message: > >/usr/lib/mingw/libmingw32.a(gccmain.o)(.text+0x88):gccmain.c: multiple >definition of `__main' >/usr/lib/libm.a(ds00023.o)(.text+0x0): first defined here If anybody already knows the cure to these problems, help me, please! thanks in advance, Janna Khegai

1 0

Pretty.lhs: bug found, feature missed
by Johannes Waldmann 09 Nov '01

09 Nov '01

We were recently bitten by what seems to be a bug in hslibs/text/Pretty.lhs. The following code prints an infinite string: import Pretty ncat x y = nest 4 $ cat [ x, y ] d1 = foldl1 ncat $ take 50 $ repeat $ char 'a' d2 = parens $ sep [ d1, text "+" , d1 ] main = print d2 I think this is due to the function indent/spaces called with a negative argument. The following is in Pretty.lhs: indent n | n >= 8 = '\t' : indent (n - 8) | otherwise = spaces n spaces 0 = "" spaces n = ' ' : spaces (n - 1) This looks Really Dangerous. Another point: we would want to use a version of `render (d :: Doc)' that always produces only a bounded amount of output, by cutting off sub-documents that are too deeply nested. (Some LISP pretty printers do have this feature - they cut long lists, and deeply nested lists). Before sitting down and trying to implement this, we would like to check whether someone already did it. If so, please tell us. Best regards, -- -- Johannes Waldmann ---- http://www.informatik.uni-leipzig.de/~joe/ -- -- joe(a)informatik.uni-leipzig.de -- phone/fax (+49) 341 9732 204/252 --

2 1

What isn't possible with Finite Maps, and should be.
by George Russell 09 Nov '01

09 Nov '01

Once again I find myself wanting something like the following function with finite maps: fmToListByRange :: Ord key => FiniteMap key elt -> (key,key) -> [(key,elt)] fmToListByRange map (k1,k2) is supposed to return all elements in the map with keys from k1 to k2 (inclusive). Although this isn't as common as looking up an element by a key, it nevertheless is sometimes useful, and shouldn't be too hard to implement, given that FiniteMaps are going to be implemented (given the Ord constraint) as some sort of tree-like query structure, so it's just a question of scanning down the tree splitting the interval as you go. Of course this can be generalised. A general range would have type (Maybe key,Maybe key,Bool) where Nothing instead of a key would mean starting at one of the ends, and the Bool would indicate whether you gave the elements in increasing or decreasing order of key. In any case it looks fairly simple to write. If only I were set up to be able to compile GHC I could write the necessary code myself . . .

2 1

RE: Sockets on Solaris
by Julian Seward (Intl Vendor) 07 Nov '01

07 Nov '01

We just tried the following prog on our Solaris build of 5.02.1: import System import Socket import IO main = do d <- connectTo "localhost" (PortNumber 7) hPutStr d "fiofofofof\n" hFlush d c <- hGetContents d putStr c and it works fine for us. The bug you report has exactly the symptoms of a bug we fixed in 5.02. You are definitely using 5.02.1, yes? J | -----Original Message----- | From: Luis Diego Fallas [mailto:luisdiegofallas@yahoo.com] | Sent: Wednesday, November 07, 2001 12:57 AM | To: glasgow-haskell-users(a)haskell.org | Subject: Sockets on Solaris | | | Hi everyone, | | I'm having trouble using the Socket module on | Solaris. | | The following program: | | import System | import Socket | import IO | | main = | do | d <- connectTo "localhost" (PortNumber 80) | hPutStr d "GET / HTTP/1.0\n\n" | hFlush d | c <- (hGetContents d) | putStr c | | | fails with: | | Fail: invalid argument | Action: getSocketOption | Reason: Invalid argument | | I'm using ghc-5.02.1 . | | Thanks. | | | | __________________________________________________ | Do You Yahoo!? | Find a job, post your resume. | http://careers.yahoo.com | | _______________________________________________ | Glasgow-haskell-users mailing list | Glasgow-haskell-users(a)haskell.org | http://www.haskell.org/mailman/listinfo/glasgo| w-haskell-users |

2 1

RE: Sockets: when and how to close them?
by Simon Marlow 07 Nov '01

07 Nov '01

> When using Sockets, exactly when and where do I have to close them? > > The documentation (of module Socket) implies that this is not > necessary? > However there is sClose in SocketPrim. My application is one Server > who runs `listenOn' and many (short lived) clients that `connectTo'. > After I run this for a while, netstat typically says > > autotool@isun11:~/autotool/serial$ netstat|grep local > localhost.9121 localhost.56749 32768 0 32768 > 0 CLOSE_WAIT > localhost.56749 localhost.9121 32768 0 32768 > 0 FIN_WAIT_2 > localhost.9121 localhost.56753 32768 0 32768 > 0 CLOSE_WAIT > localhost.56753 localhost.9121 32768 0 32768 > 0 FIN_WAIT_2 If your sockets are in the form of Handles, i.e. if you used Socket.connectTo rather than the raw version in SocketPrim, then they will be closed automatically by the standard Handle finalizer mechanism. It is good practice, however, to close them as soon as you know you no longer need them (using hClose) as this will free up the resources immediately. Sockets returned from Socket.listenOn or from the low level routines in SocketPrim must be closed by hand using SocketPrim.sClose. It is probably a bug that Socket doesn't export sClose. Cheers, Simon

1 0