October 2004 - Haskell-Cafe

IPv6?
by John Goerzen 07 Oct '04

07 Oct '04

Hi, Network.Socket in GHC6 seems to be restricted to IPv4, despite the prsence of AF_INET6. How do I support IPv6 in Haskell?

1 0

more shootout madness -- hashes in GHC
by Robert 07 Oct '04

07 Oct '04

I've been playing around with the "Hashes, Part II" code from the shootout. I wanted to try to implement this test using Data.HashTable instead of Data.FiniteMap to see if that would buy us anything. In fact, the HashTable implementation is consistantly faster than FiniteMap, but not by a lot (thus making the transition to the IO monad not worthwhile IMO). The interesting thing, however, is that at a certain number of iterations (106 in my case), the Hashtable code segfaults. GDB shows that it is blowing the top off the program stack and dying when it tries to write to kernel space. (can't write to 0xbfffffff). My question is, why does this happen? Is it well known that sequence_ ing longish lists has this effect? It seems to me that there is no reason to consume stack for sequences of IO like this (especially using sequence_ or >> where we ONLY care about the side effects of the operation). I don't have a stong grasp of how the RTS works, perhaps someone could explain in small words? The code in question is attached. As you can see, I've tried several approaches to reduce the program stack usage, but they seem to generate very similar code. -- Robert robdockins(a)fastmail.fm

1 0

Re: [Haskell-cafe] Re: OCaml list sees abysmal Language Shootout results
by MR K P SCHUPKE 07 Oct '04

07 Oct '04

I thought that you could use chained buffers to implement the lazy lists, that way the top level interface would still be simple. Also there should be an easy graph-rewrite operation that turns your 3 sequential operations (in the wc example) into 3 parallel ones... put these two together in the compiler and the simple coding should perform near to the complex form. Keean.

1 0

Re: [Haskell-cafe] Re: OCaml list sees abysmal Language Shootout
by MR K P SCHUPKE 07 Oct '04

07 Oct '04

results Okay... read the license, seems whoever wrote that article on Clean didn't read the license... You can use it under LGPL, for commercial purposes... Still cant help thinking its yet-another-lazy-functional language - I thought the whole point of Haskell was to stop prolifertion of many non-standard lazy functional languages... (have to say I like the idea of basing the formal semantics on graph rewriting though...) # Keean.

1 0

Re: [Haskell-cafe] Re: OCaml list sees abysmal Language Shootout results
by MR K P SCHUPKE 07 Oct '04

07 Oct '04

what about, quote: N.B. CleanLanguage is only available under a restrictive license. It's LGPL so long as you're not selling anything you write. Otherwise you have to purchase a commercial license. This probably means CleanLanguage will not survive long. Keean.

1 0

RE: [Haskell-cafe] What Functions are Standard?
by Simon Peyton-Jones 07 Oct '04

07 Oct '04

John | > I have been writing code using the docs over at | > http://www.haskell.org/ghc/docs/latest/html/libraries/index.html, which | > is the only comprehensive library reference I could find. | > | > I am using some code from System.IO, supposedly from base. When I try | > to build this with nhc98, it doesn't know about hGetBuf, hPutBuf, or | > openBinaryFile from there or about mallocForeignPtrArray from the | > Foreign.* area. But all these look standard to me. This is the GHC library documentation, but Hugs and NHC use the same source-code repository, but omitting some functionality. What's omitted is explicitly documented, at the exact link you give above, under "Implementation-specific notes". The language definition for Haskell 98 defines a small collection of libraries, standardised a long time ago. But all implementations come with a much richer set of libraries, many of which are written in Haskell 98, and almost all of which are usable by Haskell 98 programs. These are the ones documented above. Meanwhile lots of people are writing libraries too. In particular, there are several libraries around for reading and writing binary files, and a long recent email thread on the subject. What I'm hoping is that someone, or a small group, will decide on a good interface (in the light of experience thus far) and implement it, so that the rest of us don't have to choose. Simon

1 0

Can I determin the function name passed in?
by Mark Carter 07 Oct '04

07 Oct '04

I'm an utter noob to Haskell, and FP in gnereal. I would like to create a function taking an argument which is a function (actually an equation taking no args, but dependent on other equations), which prints the name of the function and and it's result. Is there a way of doing this in Haskell? Sounds like a job for macros, or some kind of introspection-type stuff.

4 3

Re: [Haskell-cafe] Re: OCaml list sees abysmal Language Shootout results
by MR K P SCHUPKE 07 Oct '04

07 Oct '04

Well, the sootout itself is using GHC, as it lists both language and implementation. There is no entry for NHC or Hugs... So I would say if the GHC specific code is neat and fast put it in... If people hant to see a NHC then there should be a separate implentation of haskell (for example I can see at least 3 schemes in the sootout) Keean.

2 1

Re: [Haskell-cafe] Re: OCaml list sees abysmal Language Shootout results
by MR K P SCHUPKE 07 Oct '04

07 Oct '04

looking at ghc's scores... this would seem to be the next candidate for a rework: -- $Id: reversefile.ghc.html,v 1.6 2004/10/03 00:44:58 bfulgham Exp $ -- http://www.bagley.org/~doug/shootout/ -- from Julian Assange main = interact $ unlines . reverse . lines Compare this with the gcc implementation which uses buffers and is 68 lines long! Keean.

1 0

Re: [Haskell-cafe] Re: OCaml list sees abysmal Language Shootout results
by MR K P SCHUPKE 07 Oct '04

07 Oct '04

Also, its nice to see GHC ahead of Java now in the standard results... Would be nice to beat OCaml & Clean though ... Keean.

1 0