Nabble - Haskell - Haskell-Cafe

Re: ANNOUNCE: Sun Microsystems and Haskell.org joint project on OpenSPARC

2008-07-24T04:06:45Z

On Thu, 2008-07-24 at 16:43 +1200, Richard A. O'Keefe wrote:
> On 24 Jul 2008, at 3:52 am, Duncan Coutts wrote:
> [Sun have donated a T5120 server + USD10k to develop
> support for Haskell on the SPARC.]
>
> This is wonderful news.
>
> I have a 500MHz UltraSPARC II on my desktop running Solaris 2.10.

I have 500MHz UltraSPARC II on my desktop running Gentoo Linux. :-)

> Some time ago I tried to install GHC 6.6.1 on it, but ended up
> with something that compiles to C ok, but then invokes some C
> compiler with option "-fwrapv", which no compiler on that machine
> accepts, certainly none that was present when I installed it.

I've got ghc 6.8.2 working, but only -fvia-C and only unregisterised.

"-fwrapv" is an option to some version of gcc, but I couldn't tell you
which.

> I would really love to be able to use GHC on that machine.

Me too :-), or in my case use it a bit quicker. Unregisterised ghc
builds are pretty slow.

> I also have an account on a T1 server, but the research group
> who Sun gave it to chose to run Linux on it, of all things.

Our tentative plan is to partition our T2 server using logical domains
and run both Solaris and Linux. We'd like to set up ghc build bots on
both OSs.

> So binary distributions for SPARC/Solaris and SPARC/Linux would
> be very very nice things for this new project to deliver early.

I guess this is something that anyone with an account on the box could
do. So once we get to the stage where we're handing out accounts then
hopefully this would follow.

The project isn't aiming to get the registerised C backend working
nicely, we're aiming to get a decent native backend. That should also be
much less fragile by not depending on the version of gcc so closely.

> (Or some kind of source distribution that doesn't need a working
> GHC to start with.)

That's a tad harder. Needs lot of build system hacking.

Duncan

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Re: ghc socket weirdness

2008-07-24T00:33:15Z

Tim Newsham <newsham@...> writes:

> FreeBSD/7.0 amd64:
>
> GHCi, version 6.8.2: http://www.haskell.org/ghc/ :? for help
> Loading package base ... linking ... done.
> Prelude> :module +Network
> Prelude Network> listenOn $ PortNumber 8765
> Loading package parsec-2.1.0.0 ... linking ... done.
> Loading package network-2.1.0.0 ... linking ... done.
> *** Exception: getAddrInfo: does not exist (servname not supported for
> ai_socktype)
>
> (same error when compiled).
>
> Anyone seen this before? I have multiple NICs, could that be
> confusing listenOn (does it not just bind on 0.0.0.0?)
>

ghc 6.8.2 has the same problem on NetBSD/i386, it fixed with ghc 6.8.3:

http://hackage.haskell.org/trac/ghc/ticket/2103

-- jungle
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

ghc socket weirdness

2008-07-24T00:04:21Z

FreeBSD/7.0 amd64:

GHCi, version 6.8.2: http://www.haskell.org/ghc/ :? for help
Loading package base ... linking ... done.
Prelude> :module +Network
Prelude Network> listenOn $ PortNumber 8765
Loading package parsec-2.1.0.0 ... linking ... done.
Loading package network-2.1.0.0 ... linking ... done.
*** Exception: getAddrInfo: does not exist (servname not supported for
ai_socktype)

(same error when compiled).

Anyone seen this before? I have multiple NICs, could that be
confusing listenOn (does it not just bind on 0.0.0.0?)

Tim Newsham
http://www.thenewsh.com/~newsham/
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Re: ANNOUNCE: Sun Microsystems and Haskell.org joint project on OpenSPARC

2008-07-23T22:03:01Z

On 2008 Jul 24, at 0:43, Richard A. O'Keefe wrote:

> So binary distributions for SPARC/Solaris and SPARC/Linux would
> be very very nice things for this new project to deliver early.
> (Or some kind of source distribution that doesn't need a working
> GHC to start with.

I'm still working on SPARC/Solaris here as well. (Still trying to get
a build that doesn't produce executables that throw "schedule re-
entered unsafely" immediately on startup.)

--
brandon s. allbery [solaris,freebsd,perl,pugs,haskell] allbery@...
system administrator [openafs,heimdal,too many hats] allbery@...
electrical and computer engineering, carnegie mellon university KF8NH

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Re: ANNOUNCE: Sun Microsystems and Haskell.org joint project on OpenSPARC

2008-07-23T21:43:59Z

On 24 Jul 2008, at 3:52 am, Duncan Coutts wrote:
[Sun have donated a T5120 server + USD10k to develop
support for Haskell on the SPARC.]

This is wonderful news.

I have a 500MHz UltraSPARC II on my desktop running Solaris 2.10.
Some time ago I tried to install GHC 6.6.1 on it, but ended up
with something that compiles to C ok, but then invokes some C
compiler with option "-fwrapv", which no compiler on that machine
accepts, certainly none that was present when I installed it.

I would really love to be able to use GHC on that machine.

I also have an account on a T1 server, but the research group
who Sun gave it to chose to run Linux on it, of all things.

So binary distributions for SPARC/Solaris and SPARC/Linux would
be very very nice things for this new project to deliver early.
(Or some kind of source distribution that doesn't need a working
GHC to start with.)

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Re: Literate program implementing text editor

2008-07-23T15:43:06Z

On Wed, 2008-07-23 at 21:05 +0100, Eric wrote:
> Hi all,
>
> I wanted to embark on a Haskell program that implements a simple text
> editor. Before doing so, however, I thought that I should ask: Does
> anyone know of a literate Haskell program already in existence that does
> the job?

I don't know about simple, but there is Yi and once upon a time it was
simple before everyone added lots of features to it. It's probably not
literate Haskell though.

Duncan

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Re: More fun with micro-benchmarks and optimizations. (GHC vs Perl)

2008-07-23T15:33:20Z

On Wed, Jul 23, 2008 at 6:10 PM, Corey O'Connor <coreyoconnor@...> wrote:

> My straightforward Haskell solution was:
> -------------------------------------------------------------------
> import Text.Regex.Posix
>
> main = do
> f_lines <- readFile "test.out" >>= return . lines
> let duration = foldl add_line 0.0 f_lines
> add_line sum line =
> let [[_,m]] = line =~ "([0-9.]+)"
> framerate = read m
> delta = 1.0 / framerate
> in sum + delta
> putStrLn $ "Duration (sec): " ++ show duration
> -------------------------------------------------------------------

I'm sure the other suggestions have improved performance far more than
this suggestion will, but it's still worth a shot.

I don't think I've _ever_ seen a case where using foldl was a good
idea. foldl interacts poorly with laziness; use foldr when the
function you are folding is nonstrict, and use foldl' (from
Data.List), a strict version of foldl, when the function you are
folding is strict. This is, of course, only a rule of thumb.

Changing that foldl to foldl' will probably speed this up a decent amount.

Luke
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Re: BLAS Solve Example

2008-07-23T15:30:16Z

A jedi master might stick with the existing double precision solver,
then convert the results to best rational approximation [1], then do a
forward solve on the rational versions of matrices, adjusting numerator
and denominator to eliminate the residual error (with a heuristic to
favor common factors). If you are very lucky, such a rational number
will exist, depending on your limits of humongous.

[1] e.g. http://www.dtashley.com/howtos/2007/01/best_rational_approximation/

Darrin Thompson wrote:

> On Wed, Jul 23, 2008 at 2:12 AM, Alberto Ruiz <aruiz@...> wrote:
>> $ ghci solve.hs
>> *Main> sol
>> 3 |> [-5.555555555555511e-2,0.11111111111111113,0.2777777777777776]
>>
>
> I was hoping for rational solutions. If I were a true jedi master I'd
> write my own solver, which might be the right thing to do. All I know
> so far is gauss' method. Probably I'd learn something implementing the
> back substitution. hmm....
>
> Thanks.
>
> --
> Darrin
> _______________________________________________
> Haskell-Cafe mailing list
> Haskell-Cafe@...
> http://www.haskell.org/mailman/listinfo/haskell-cafe
>
>

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Re: [Off-topic] Loss of humour

2008-07-23T14:12:57Z

On 24 Jul 2008, at 00:45, Andrew Coppin wrote:

> Miguel Mitrofanov wrote:
>>> Also... the current Humour page on the Haskell wiki contains a
>>> link to Lambdabot's quotes database, but on my system, clicking
>>> this link just displays a few hundred pages of gibberish. Is this
>>> normal?
>>
>> Of course it's not. But deciphering is very simple, it's named
>> "gunzip".
>
> Right... so why does it have the MIME type "text/plain"? o_O

Probably because it should be in "Content-Encoding" header.

See, most web browsers are capable of gunzipping content - but only if
they know it's gzipped, so they look at the Content-Encoding header,
and if it's present and says "gzip", it decompresses the content. This
is perfectly safe, since before receiving content browser lets the
server know it accepts compressed pages (by sending "Accept-Encoding"
header), so the server knows it can send compressed content.

> (Also... how do you know it's ZIP, and not, say, gzip, bz2, RAR,
> LHA...? I can't see anything that looks like an identifying header
> in there.)

Hmmm... your "file" tool doesn't know about gzip archives?
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Re: [Off-topic] Loss of humour

2008-07-23T14:09:08Z

Jeremy Shaw wrote:

> At Wed, 23 Jul 2008 19:45:56 +0100,
> Andrew Coppin wrote:
>
>> A while back I found a page somewhere containing some rather amusing IRC
>> quotes. Unfortunately it seems to have vanished. I can't remember where
>> on earth I found it, but I've scoured the Internet trying to track it
>> down. (In particular, it contained a quote of somebody impersonating a
>> typical Haskell newbie - lots of enthusiasm and no attention span! Well
>> it amused *me* anyway...) Anybody have any ideas where this has gone?
>>
>
> http://web.archive.org/web/20070609061216/http://www.haskell.org/hawiki/QuotesPage
>

LOL!

Thanks, that's the one...

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Re: [Off-topic] Loss of humour

2008-07-23T14:00:50Z

http://bash.org ?

On Wed, Jul 23, 2008 at 2:45 PM, Andrew Coppin
<andrewcoppin@...> wrote:

> A while back I found a page somewhere containing some rather amusing IRC
> quotes. Unfortunately it seems to have vanished. I can't remember where on
> earth I found it, but I've scoured the Internet trying to track it down. (In
> particular, it contained a quote of somebody impersonating a typical Haskell
> newbie - lots of enthusiasm and no attention span! Well it amused *me*
> anyway...) Anybody have any ideas where this has gone?
>
> Also... the current Humour page on the Haskell wiki contains a link to
> Lambdabot's quotes database, but on my system, clicking this link just
> displays a few hundred pages of gibberish. Is this normal?
>
> As I final note...
> http://www.haskell.org/haskellwiki/Humor/Goldilocks
> Damn, I wish *I* thought of that! ;-)
>
> _______________________________________________
> Haskell-Cafe mailing list
> Haskell-Cafe@...
> http://www.haskell.org/mailman/listinfo/haskell-cafe
>

--
I try to take things like a crow; war and chaos don't always ruin a
picnic, they just mean you have to be careful what you swallow.

-- Jessica Edwards
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Re: More fun with micro-benchmarks and optimizations. (GHC vs Perl)

2008-07-23T14:00:12Z

brad.larsen:

> And against gawk 3.1.5:
>
> $ time awk -F: '{sum += 1 / $2} END{print sum}' test.out
> 3155.63
>
> real 0m0.197s
> user 0m0.184s
> sys 0m0.004s
>
> compared to Don's Haskell version:
>
> $ time ./fastSum < test.out
> 3155.626666664377
>
> real 0m0.072s
> user 0m0.056s
> sys 0m0.004s
>
> compared to the Corey's perl version:
>
> $ time perl Sum.pl
> Duration (sec): 3155.62666666438
>
> real 0m0.181s
> user 0m0.164s
> sys 0m0.012s
>

Another variant, using a non-copying readDouble (available in
bytestring-lexing package 0.1.2),

{-# OPTIONS -fbang-patterns #-}

import qualified Data.ByteString.Char8 as S
import qualified Data.ByteString.Unsafe as S
import Data.ByteString.Lex.Double

main = print . go 0 =<< S.getContents
where
go :: Double -> S.ByteString -> Double
go !n !s = case unsafeReadDouble s' of
Nothing -> n
Just (k,t) -> let delta = 1 / k in go (n+delta) t
where
s' = case S.elemIndex ':' s of
Nothing -> S.empty
Just i -> S.unsafeDrop (i+2) s

Computes in:

$ time ./Fast < test.out
3155.626666664377
./Fast < test.out 0.02s user 0.00s system 80% cpu 0.025 total

So that's probably a good place to stop.

Note the general rules:

* strict bytestrings
* tight tail-recursive loops with strict, atomic accumulators
* non-copying parsing.

-- Don
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Re: [Off-topic] Loss of humour

2008-07-23T13:57:37Z

At Wed, 23 Jul 2008 19:45:56 +0100,
Andrew Coppin wrote:
>
> A while back I found a page somewhere containing some rather amusing IRC
> quotes. Unfortunately it seems to have vanished. I can't remember where
> on earth I found it, but I've scoured the Internet trying to track it
> down. (In particular, it contained a quote of somebody impersonating a
> typical Haskell newbie - lots of enthusiasm and no attention span! Well
> it amused *me* anyway...) Anybody have any ideas where this has gone?

http://web.archive.org/web/20070609061216/http://www.haskell.org/hawiki/QuotesPage

j.
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

RE: Re: Aren't type system extensions fun? [Further analysis]

2008-07-23T13:47:06Z

Andrew writes

| I have complete confidence that whoever wrote the GHC manual knew
| exactly what they meant. I am also fairly confident that this was the
| same person who implemented and even designed this particular feature.
| And that they probably have an advanced degree in type system theory. I,
| however, do not. So if in future the GHC manual could explain this in
| less opaque language, that would be most appreciated. :-}

As the person who wrote the offending section of the GHC user manual (albeit lacking a qualification of any sort in type theory), I am painfully aware of its shortcomings. Much of the manual is imprecise, and explains things only by examples. The wonder is that it apparently serves the purpose for many people.

Nevertheless, the fact is that it didn't do the job for Andrew, at least not first time around.

After some useful replies to his email, he wrote

| Thank you for coming up with a clear and comprehensible answer.

I would very much welcome anyone who felt able to improve on the text in the manual, even if it's only a paragraph or two. If you stub your toe on the manual, once you find the solution, take a few minutes to write the words you wish you had read to begin with, and send them to me.

Meanwhile, Andrew, I recommend the paper "Practical type inference for arbitrary rank types", which is, compared to the user manual at least, rather carefully written.

Simon
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

RE: Re: Profiling nested case

2008-07-23T13:47:00Z

| I had similar experiences as you when attempting to write "high
| performance Haskell"; the language makes you want to use high-level
| abstracted functions but the optimizer (while amazing, to be honest)
| seems to miss a few cases that it seems like it should hit.

Ryan, if you find any of these, do please submit them to GHC's Trac bug-tracker. There's a special category for performance bugs. Small programs of the form "GHC should hit this, but doesn't" are incredibly useful.

Thanks

Simon
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Re: [Off-topic] Loss of humour

2008-07-23T13:45:45Z

Miguel Mitrofanov wrote:
>> Also... the current Humour page on the Haskell wiki contains a link
>> to Lambdabot's quotes database, but on my system, clicking this link
>> just displays a few hundred pages of gibberish. Is this normal?
>
> Of course it's not. But deciphering is very simple, it's named "gunzip".

Right... so why does it have the MIME type "text/plain"? o_O

(Also... how do you know it's ZIP, and not, say, gzip, bz2, RAR, LHA...?
I can't see anything that looks like an identifying header in there.)

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Re: [Off-topic] Loss of humour

2008-07-23T13:35:43Z

> Also... the current Humour page on the Haskell wiki contains a link
> to Lambdabot's quotes database, but on my system, clicking this link
> just displays a few hundred pages of gibberish. Is this normal?

Of course it's not. But deciphering is very simple, it's named "gunzip".

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Re: [Off-topic] Loss of humour

2008-07-23T13:29:31Z

On Jul 23, 2008, at 1:45 PM, Andrew Coppin wrote:
> A while back I found a page somewhere containing some rather amusing
> IRC quotes.

Are you perhaps thinking of the "Quotes of the Week" section in the
Haskell Weekly News?

Back issues seem to be at http://sequence.complete.org/hwn if you want
to check.

-johnnnnnn

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Haskell Weekly News: Issue 78 - July 23, 2008

2008-07-23T13:13:45Z

---------------------------------------------------------------------------
Haskell Weekly News
http://sequence.complete.org/hwn/20080723
Issue 78 - July 23, 2008
---------------------------------------------------------------------------

Welcome to issue 78 of HWN, a newsletter covering developments in the
[1]Haskell community.

Issue 78: In Which Michi and Neil Become Doctors, Sun Donates Some
Sweet Loot, and Jules Is Revealed To Be A Helpful Anthropomorphic Robot

Community News

Congratulations are in order this week to two members of the community
who have completed PhDs. Neil Mitchell (ndm) passed his PhD viva last
week, subject to minor corrections. Mikael Johansson (Syzygy-) [2]has
also completed his PhD and will soon be starting a postdoc at Stanford
with the topology in computer science working group. Congratulations,
Drs. Mitchell and Johansson!

Announcements

rosezipper. Eric Kow (kowey) [3]announced the release of [4]rosezipper,
Krasimir Angelov and Iavor S. Diatchki's Data.Tree implementation of
zippers.

list-extras 0.1.0. wren ng thornton (koninkje) [5]announced the initial
release of [6]list-extras, a home for common not-so-common list
functions.

Sun Microsystems and Haskell.org joint project on OpenSPARC. Duncan
Coutts (dcoutts) [7]announced a [8]joint project between Sun
Microsystems and the Haskell.org community to exploit the high
performance capabilities of Sun's latest multi-core [9]OpenSPARC
systems via Haskell! Sun has donated a powerful 8 core SPARC Enterprise
T5120 Server to the Haskell community, and $10,000 to fund a student
to further develop support for high performance Haskell on the SPARC.
The student will work with a mentor from Haskell.org and an adviser
from Sun's SPARC compiler team. If you're a student and this sounds
interesting to you, send in those applications!!

Hayoo! beta 0.2. Timo B. [10]announced the second beta release of
[11]Hayoo!, a Haskell API search engine providing advanced features
like suggestions, find-as-you-type, fuzzy queries and much more. The
major change in this release is the inclusion of all packages available
on Hackage in the index.

Haskell-beginners mailing list. Benjamin L. Russell [12]announced the
creation of the [13]Haskell-Beginners Mailing List, beginners at
haskell.org, devoted to discussion of primarily beginner-level topics
related to Haskell. It's already off to a great start, so if you're a
Haskell beginner, or someone interested in answering beginner
questions, please subscribe!

Haskeline 0.2. Judah Jacobson [14]announced the initial (alpha-ish)
release of [15]Haskeline, a library for line input in command-line
programs. It is similar in purpose to editline or readline, but is
written in Haskell and thus (hopefully) more easily used in other
Haskell programs.

Google Summer of Code

Progress updates from participants in the 2008 [16]Google Summer of
Code.

Generic tries. Jamie Brandon is working on a library for efficient maps
using generalized tries. [17]This week, he ran QuickCheck on his test
suite for the first time, and found a large number of failing tests!
He's got his work cut out for him straightening those out over the next
few days.

DPH physics engine. Roman Cheplyaka (Feuerbach) is working on a
[18]physics engine using [19]Data Parallel Haskell. [20]This week, he
implemented full handling of rigid body collisions, including angular
velocity. Next he plans to explore various ways to make the engine
faster, including broad-phase collision detection.

GHC plugins. Max Bolingbroke is working on dynamically loaded plugins
for GHC. [21]This week, he revealed his "mystery project": an HTML
pretty-printer for GHC core! [22]Here is a sample. Now his focus turns
to tidying things up and solidifying documentation in preparation for
getting his patches merged into GHC HEAD.

Hoogle 4. Neil Mitchell (ndm) is working on [23]Hoogle 4. [24]This
week, he fleshed out the final part of type search, including support
for instances and alpha renaming of variables. Unfortunately, it uses
too much memory to be feasibly run on the base libraries! Neil has some
ideas on how to fix this, however, which he plans to tackle next week.

Language.C. Benedikt Huber (visq) is [25]working on Language.C, a
standalone parser/pretty printer library for C99. He has finally
completed a working implementation for analysing declarations and
definitions, and [26]presents a working example of the library's use.

Cabal dependency framework. Andrea Vezzosi (Saizan) is working on a
[27]make-like dependency analysis framework for Cabal.

GHC API. Thomas Schilling (nominolo) is working on [28]improvements to
the GHC API.

Libraries

Proposals and extensions to the [29]standard libraries.

generalize mapAccumL/R. Ross Paterson [30]proposed adding generalized
versions of mapAccumL and mapAccumR to Data.Traversable.

signal-handling API. Simon Marlow [31]proposed a [32]new
signal-handling API.

Discussion

Point-free style in guards. L29Ah [33]asked a question about using a
points-free style in guard expressions, leading to a number of clever
suggestions involving custom combinators.

Optimizing sequence. Gracjan Polak started a [34]discussion on the
strictness properties of the sequence function, and its implications
for optimization. It sounds as though adding an alternate strict
version of sequence to the libraries could be a good idea.

Blog noise

[35]Haskell news from the [36]blogosphere.

* Eric Kow (kowey): [37]rose zipper on hackage. Eric has uploaded an
implementation of a zipper for Data.Tree to hackage.
* Jamie Brandon: [38]QuickCheck strikes again. An update on Jamie's
Google Summer of Code project. Lots of failing QuickCheck tests
mean Jamie has his work cut out for him.
* >>> Ayumilove: [39]Haskell Programming Tutorial Part 1. A video
introduction to Haskell?
* Roman Cheplyaka (Feuerbach): [40]Status report: week 7-8. An update
on Roman's Google Summer of Code project.
* Conal Elliott (conal): [41]Designing for the future.
* Benedikt Huber (visq): [42]Language.C: Analysing Definitions. An
update on Benedikt's Google Summer of Code project.
* Max Bolingbroke: [43]Compiler Plugins For GHC: Week Six. An update
on Max's Google Summer of Code project.
* Arnar Birgisson (Arnar): [44]Playing with Haskell's lazy lists.
Arnar implements Eric Rowland's [45]simple prime-generating
recurrence using lazy lists in Haskell.
* Neil Mitchell (ndm): [46]GSoC Hoogle: Week 8. An update on Neil's
Google Summer of Code project.
* Matthew Sackman: [47]What is the point?. Matthew had his submission
rejected from the Haskell Symposium, and isn't happy about it.
* Tupil: [48]Stemming with Haskell reloaded. An updated, more
functional/Haskellish interface to the stemmer library.
* Luke Palmer (luqui): [49]Semantic Design. Luke's reflections on a
design approach learned from Conal Elliott.
* Muad`Dib (vixey): [50]Rascal - Mini-haskell like language.
* Jeremy Shaw: [51]HTML Templating in HAppS using HSP. Jeremy writes
a very cool introduction/tutorial to using HSP (Haskell Server
Pages) with HAppS.
* >>> Duane Johnson: [52]Lazy Evaluation at Work. Duane likes the
idea of lazy file I/O.
* >>> Duane Johnson: [53]A Glimmer of Monadic Hope. Duane figures out
do-notation.
* >>> Duane Johnson: [54]Using 'foldr' in Haskell. Duane figures out
how to define (++) and concat in terms of foldr.
* Mikael Johansson (Syzygy-): [55]Dr rer nat, Magna cum laude.
* >>> Holden Karau: [56]Integrating your HUnit (or other) tests into
your cabal package.
* Real-World Haskell: [57]Beta availability hiccups.
* >>> Antoine Hersen: [58]ICFP 2008 Postmortem.
* Edward Kmett (edwardk): [59]A Sort of Difference. Edward uses an
analysis of quicksort in Haskell as a jumping-off point for an
introduction to difference lists.
* Ralf Lammel: [60]The Expression Lemma -- Explained. Ralf explains
the relationship between OOP and FP.
* >>> Greg McClure: [61]Learning languages through problem-solving.
Greg extols the virtues of [62]Project Euler for learning new
languages, and exhibits a solution to the first problem in Haskell,
Erlang, and Python.

Quotes of the Week

* kryptiskt: my point is that our brain isn't some logic machine,
it's a jury-rigged contraption to help us get food, friends and
sex. It's the Perl of intelligences.
* Fallacy: peyton `simon` jones
* Quadrescence: [on #haskell] Well, the nice population here attracts
people. Unlike other pla_C_es. How'd those underscores get there?
* mauke: hmm, regexes get even more cryptic after z-encoding:
ZLz3fUZCZLz3fUzlznzrwZRZLz3fUzezrwZRzbZLz3fUzlzezrwZRZLz3fUznzrwZRZR
* BMeph: Haskell: Where even the newest newcomer acts monadically:
join :: ask (ask something) -> ask something
* Quadrescence: quicksilver: You must be an anthropomorphic robot or
something.

About the Haskell Weekly News

New editions are posted to [63]the Haskell mailing list as well as to
[64]the Haskell Sequence and [65]Planet Haskell. [66]RSS is also
available, and headlines appear on [67]haskell.org. Headlines are
available as [68]PDF.

To help create new editions of this newsletter, please see the
information on [69]how to contribute. Send stories to byorgey at seas
dot upenn dot edu. The darcs repository is available at darcs get
[70]http://code.haskell.org/~byorgey/code/hwn/ .

References

1. http://haskell.org/
2. http://blog.mikael.johanssons.org/archive/2008/07/dr-rer-nat-magna-cum-laude/
3. http://koweycode.blogspot.com/2008/07/rose-zipper-on-hackage.html
4. http://hackage.haskell.org/cgi-bin/hackage-scripts/package/rosezipper
5. http://article.gmane.org/gmane.comp.lang.haskell.cafe/42485
6. http://code.haskell.org/~wren/
7. http://article.gmane.org/gmane.comp.lang.haskell.general/16351
8. http://haskell.org/opensparc/
9. http://opensparc.net/
10. http://article.gmane.org/gmane.comp.lang.haskell.general/16349
11. http://holumbus.fh-wedel.de/hayoo
12. http://article.gmane.org/gmane.comp.lang.haskell.general/16345
13. http://www.haskell.org/mailman/listinfo/beginners
14. http://article.gmane.org/gmane.comp.lang.haskell.general/16343
15. http://hackage.haskell.org/cgi-bin/hackage-scripts/package/haskeline
16. http://hackage.haskell.org/trac/summer-of-code/wiki/SoC2008
17. http://jamiiecb.blogspot.com/2008/07/quickcheck-strikes-again.html
18. http://haskell.org/haskellwiki/Hpysics
19. http://www.haskell.org/haskellwiki/GHC/Data_Parallel_Haskell
20. http://physics-dph.blogspot.com/2008/07/status-report-week-7-8.html
21. http://blog.omega-prime.co.uk/2008/07/21/compiler-plugins-for-ghc-week-six/
22. http://www.omega-prime.co.uk/files/GHC-Core-HTML.html
23. http://code.haskell.org/hoogle/
24. http://neilmitchell.blogspot.com/2008/07/gsoc-hoogle-week-8.html
25. http://www.sivity.net/projects/language.c/wiki/
26. http://hsbene.blogspot.com/2008/07/languagec-analysing-definitions.html
27. http://code.haskell.org/~Saizan/cabal
28. http://hackage.haskell.org/trac/ghc/wiki/GhcApiStatus
29. http://haskell.org/haskellwiki/Library_submissions
30. http://article.gmane.org/gmane.comp.lang.haskell.libraries/9649
31. http://article.gmane.org/gmane.comp.lang.haskell.libraries/9599
32. http://hackage.haskell.org/trac/ghc/ticket/2451
33. http://thread.gmane.org/gmane.comp.lang.haskell.cafe/42475
34. http://thread.gmane.org/gmane.comp.lang.haskell.cafe/42453
35. http://planet.haskell.org/
36. http://haskell.org/haskellwiki/Blog_articles
37. http://koweycode.blogspot.com/2008/07/rose-zipper-on-hackage.html
38. http://jamiiecb.blogspot.com/2008/07/quickcheck-strikes-again.html
39. http://www.5min.com/Video/Ayumilove-Haskell-Programming-Tutorial-Part-1-31180892
40. http://physics-dph.blogspot.com/2008/07/status-report-week-7-8.html
41. http://feeds.feedburner.com/~r/conal/~3/342784118/
42. http://hsbene.blogspot.com/2008/07/languagec-analysing-definitions.html
43. http://blog.omega-prime.co.uk/2008/07/21/compiler-plugins-for-ghc-week-six/
44. http://www.hvergi.net/2008/07/playing-with-haskells-lazy-lists/
45. http://recursed.blogspot.com/2008/07/rutgers-graduate-student-finds-new.html
46. http://neilmitchell.blogspot.com/2008/07/gsoc-hoogle-week-8.html
47. http://www.wellquite.org/what_is_the_point.html
48. http://blog.tupil.com/stemming-with-haskell-reloaded/
49. http://luqui.org/blog/archives/2008/07/18/semantic-design/
50. http://muaddibspace.blogspot.com/2008/07/rascal-mini-haskell-like-language.html
51. http://nhlab.blogspot.com/2008/07/html-templating-in-happs-using-hsp.html
52. http://blog.inquirylabs.com/2008/07/18/lazy-evaluation-at-work/
53. http://blog.inquirylabs.com/2008/07/17/a-glimmer-of-monadic-hope/
54. http://blog.inquirylabs.com/2008/07/17/using-foldr-in-haskell/
55. http://blog.mikael.johanssons.org/archive/2008/07/dr-rer-nat-magna-cum-laude/
56. http://feeds.feedburner.com/~r/holdenkarau/iYtm/~3/337908379/integrating-your-hunit-or-other-tests.html
57. http://www.realworldhaskell.org/blog/2008/07/16/beta-availability-hiccups/
58. http://endymion2021.livejournal.com/17117.html
59. http://comonad.com/reader/2008/a-sort-of-difference/
60. http://blogs.msdn.com/ralflammel/archive/2008/07/16/the-expression-lemma-explained.aspx
61. http://www.codechops.com/?p=15
62. http://www.projecteuler.net/
63. http://www.haskell.org/mailman/listinfo/haskell
64. http://sequence.complete.org/
65. http://planet.haskell.org/
66. http://sequence.complete.org/node/feed
67. http://haskell.org/
68. http://code.haskell.org/~byorgey/code/hwn/archives/20080723.pdf
69. http://haskell.org/haskellwiki/HWN
70. http://code.haskell.org/~byorgey/code/hwn/
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Literate program implementing text editor

2008-07-23T13:05:39Z

Hi all,

I wanted to embark on a Haskell program that implements a simple text
editor. Before doing so, however, I thought that I should ask: Does
anyone know of a literate Haskell program already in existence that does
the job?

Eric M.

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Re: Re: Optimizing 'sequence'

2008-07-23T12:52:45Z

Janis Voigtlaender wrote:
> How about some QuickChecking in connection with the "Chasing bottoms"
> library (http://citeseer.ist.psu.edu/704350.html)?

Ah, finally a reference to what this curios phrase is actually about...!

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Re: More fun with micro-benchmarks and optimizations. (GHC vs Perl)

2008-07-23T12:37:08Z

On Wed, Jul 23, 2008 at 2:23 PM, Justin Bailey <jgbailey@...> wrote:

> On Wed, Jul 23, 2008 at 11:10 AM, Corey O'Connor <coreyoconnor@...> wrote:
>> I still have two questions after all this:
>> - Can I get a Haskell implementation as fast as the Perl?
>> - What do I need to do to get GHC's profiler to provide me usable
>> information? Telling me that 98% of the time was in "main" is not very
>> enlightening ;-)
>
> Don did a great job answering teh first question. To get better
> profiling information, add "SCC" annotations to your code. That will
> tell you how much time is spent in a particular expression. For your
> main function, take a divide and conquer approach - add SCCs to
> determine what is taking the most time.

I did write a version of SumTiny.hs with SCC annotations:
http://tothepowerofdisco.com/repo/sum_optimization/SumTinyProf.hs
Maybe I'm missing something, but the profiling still gives me poor
results. Here is the results from the profiler (cut down and, no
doubt, with poor post-email formatting):
-------------------------------------------------------------------
COST CENTRE %time %alloc

do 92.8 87.6
regex 2.8 4.8
add_line 2.2 0.4
div 1.1 0.3
readFile 0.6 6.2

individual inherited
COST CENTRE %time %alloc %time %alloc

MAIN
CAF 0.0 0.0 100.0 100.0
main 0.0 0.0 100.0 100.0
do 92.8 87.6 100.0 100.0
show 0.0 0.0 0.0 0.0
foldl 0.0 0.3 0.0 0.3
add_line 2.2 0.4 6.6 5.9
regex 2.8 4.8 2.8 4.8
sum 0.0 0.1 0.0 0.1
div 1.1 0.3 1.1 0.3
read Double 0.6 0.2 0.6 0.2
readFile 0.6 6.2 0.6 6.2
-------------------------------------------------------------------

Which tells me the "do" expression that main is equal to is
responsible for 92.8% of the run time. Which really doesn't tell me
anything.

--
-Corey O'Connor
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Re: More fun with micro-benchmarks and optimizations. (GHC vs Perl)

2008-07-23T12:37:00Z

And against gawk 3.1.5:

$ time awk -F: '{sum += 1 / $2} END{print sum}' test.out
3155.63

real 0m0.197s
user 0m0.184s
sys 0m0.004s

compared to Don's Haskell version:

$ time ./fastSum < test.out
3155.626666664377

real 0m0.072s
user 0m0.056s
sys 0m0.004s

compared to the Corey's perl version:

$ time perl Sum.pl
Duration (sec): 3155.62666666438

real 0m0.181s
user 0m0.164s
sys 0m0.012s

Regards,
Brad Larsen

On Wed, 23 Jul 2008 15:01:24 -0400, Don Stewart <dons@...> wrote:

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Libevent FFI problems

2008-07-23T12:28:32Z

Dear list members,

This is my first attempt to create a FFI to libevent
(http://monkey.org/~provos/libevent/) which is an event notification
library. A simple usage example is for instance given here:
http://unx.ca/log/libevent_echosrv1c/ . In C one basically creates
struct event instances which are initialized with a callback function
and the kind of event (e.g. read) one is intested in. Optionally a
data argument can be given which will be passed to the callback
function when invoked eventually. An event dispatcher loop will invoke
the callback when the event occurs.

I have attached my first attempt at a FFI to libevent. However it
doesn't work correctly. In the client code I want to pass 99 as data
to the callback function, but what is printed out is 0. I guess there
is an issue with pointer derefencing, but I don't know.

Also I am not sure how to handle the FunPtr deallocation. Since the
FunPtr is stored in the foreign event structure I don't know when I
can free it.

I would be grateful for any advices, hints or comments. And I really
look forward to the FFI section in the "Real World Haskell" book.

Thank you very much.

- Levi

[main.hs]

module Main where import System.IO import Network.Socket import Network.Stream import Network.HTTP import Network.URI import Control.Monad import Control.Exception import Control.Concurrent import System.Posix.IO import System.Posix.Types import Event as E import qualified Data.Map as Map callback :: Fd -> E.EventType -> Int -> IO () callback fd e x = do print fd print e print x main :: IO () main = withSocketsDo $ do E.init addrinfos <- getAddrInfo (Just (defaultHints {addrFlags = [AI_PASSIVE]})) Nothing (Just "1234") let serveraddr = head addrinfos -- create server socket sock <- socket (addrFamily serveraddr) Stream defaultProtocol -- bind it to the address we're listening to bindSocket sock (addrAddress serveraddr) -- start listening for connection requests -- max. connection requests waiting to be accepted = 5 (max. queue size) listen sock 5 ev <- E.createEvent (Fd (fdSocket sock)) [E.ReadEvent, E.WriteEvent] callback 99 E.addEvent ev Nothing E.dispatch return ()

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

hsevent.hsc (4K) Download Attachment

Re: More fun with micro-benchmarks and optimizations. (GHC vs Perl)

2008-07-23T12:23:48Z

On Wed, Jul 23, 2008 at 11:10 AM, Corey O'Connor <coreyoconnor@...> wrote:
> I still have two questions after all this:
> - Can I get a Haskell implementation as fast as the Perl?
> - What do I need to do to get GHC's profiler to provide me usable
> information? Telling me that 98% of the time was in "main" is not very
> enlightening ;-)

Don did a great job answering teh first question. To get better
profiling information, add "SCC" annotations to your code. That will
tell you how much time is spent in a particular expression. For your
main function, take a divide and conquer approach - add SCCs to
determine what is taking the most time.

Take Don's advice first though before benchmarking - you probably
won't care after switching to bytestring & a better regex engine.

Justin
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Re: More fun with micro-benchmarks and optimizations. (GHC vs Perl)

2008-07-23T12:01:24Z

coreyoconnor:

{-# OPTIONS -fbang-patterns #-}

import qualified Data.ByteString.Char8 as S
import Data.ByteString.Lex.Double
import Debug.Trace

main = print . go 0 =<< S.getContents
where
go !n !s = case readDouble str of
Nothing -> n
Just (k,t) -> let delta = 1.0 / k in go (n+delta) t
where
(_, xs) = S.break ((==) ':') s
str = S.drop 2 xs

It uses the bytestring-lexing package on Hackage to read the Doubles
out,

$ ghc --make Fast.hs -O2 -fvia-C -optc-O2
$ time ./Fast < test.out
3155.626666664377
./Fast < test.out 0.07s user 0.01s system 97% cpu 0.078 total

So that's twice as fast as the perl entry on my box,

$ time perl Sum.pl < test.out
Duration (sec): 3155.62666666438
perl Sum.pl < test.out 0.15s user 0.03s system 100% cpu 0.180 total

Note that the safe Double lexer uses a bit of copying, so
we can in fact do better still with a non-copying Double parser,
but that's only for the hardcore.

-- Don
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

InterleavableIO

2008-07-23T11:50:33Z

Hello,

I've done a package based on the Jules Bean's e-mail [1] to
Haskell-Café, with a more generalized implementation. In the original
version, you could use InterleavableIO with StateT Int IO (), but not
with StateT Int (ReaderT Char IO) (). In this version, this is
possible. The package, interleavableIO, is available at [2]. I've also
done an application that generates a version of a module with
generalized parameters, called interleavableGen, available at [3]. This
is very alpha, and interleavableGen has been tested only with
Control.Exception yet.

Comments are very welcome.

Greetings.

[1]: http://www.haskell.org/pipermail/haskell-cafe/2007-July/028501.html
[2]:
http://hackage.haskell.org/cgi-bin/hackage-scripts/package/interleavableIO-0.0.1
[3]:
http://hackage.haskell.org/cgi-bin/hackage-scripts/package/interleavableGen-0.0.1

--
Marco Túlio Gontijo e Silva
Página: http://marcotmarcot.googlepages.com/
Blog: http://marcotmarcot.blogspot.com/
Correio: marcot@...
XMPP: marcot@...
IRC: marcot@...
Telefone: 25151920
Celular: 98116720
Endereço:
Rua Turfa, 639/701
Prado 30410-370
Belo Horizonte/MG Brasil

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

[Off-topic] Loss of humour

2008-07-23T11:45:56Z

A while back I found a page somewhere containing some rather amusing IRC
quotes. Unfortunately it seems to have vanished. I can't remember where
on earth I found it, but I've scoured the Internet trying to track it
down. (In particular, it contained a quote of somebody impersonating a
typical Haskell newbie - lots of enthusiasm and no attention span! Well
it amused *me* anyway...) Anybody have any ideas where this has gone?

Also... the current Humour page on the Haskell wiki contains a link to
Lambdabot's quotes database, but on my system, clicking this link just
displays a few hundred pages of gibberish. Is this normal?

As I final note...
http://www.haskell.org/haskellwiki/Humor/Goldilocks
Damn, I wish *I* thought of that! ;-)

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Re: More fun with micro-benchmarks and optimizations. (GHC vs Perl)

2008-07-23T11:26:51Z

Sounds great! Thanks for the advice. :-)

-Corey

On Wed, Jul 23, 2008 at 1:18 PM, Don Stewart <dons@...> wrote:

> coreyoconnor:
>> I have the need to regularly write tiny programs that analyze output
>> logs. The output logs don't have a consistent formatting so I
>> typically choose Perl for these tasks.
>>
>> The latest instance of having to write such a program was simple
>> enough I figured I'd try my hand at using Haskell instead. The short
>> story is that I could quickly write a Haskell program that achieved
>> the goal. Yay! However, the performance was ~8x slower than a
>> comparable Perl implementation. With a lot of effort I got the Haskell
>> version to only 2x slower. A lot of the optimization was done with
>> guesses that the performance difference was due to how each line was
>> being read from the file. I couldn't determine much using GHC's
>> profiler.
>>
>> I still have two questions after all this:
>> - Can I get a Haskell implementation as fast as the Perl?
>> - What do I need to do to get GHC's profiler to provide me usable
>> information? Telling me that 98% of the time was in "main" is not very
>> enlightening ;-)
>>
>> All the code and data for this little experiment I've placed here:
>> http://tothepowerofdisco.com/repo/sum_optimization/
>> My first, and shortest, version is SumTiny.hs. The optimized version
>> is SumFast.hs.
>>
>> The long version for the curious:
>>
>> The (cleaned up) data was a 78k line file consisting of lines like the
>> following:
>> framerate (prev == no pts): 15
>> framerate (delta): 25
>> framerate (invalid timebase): 12.5
>> ... and so on.
>>
>> The need was for a program that calculated 1.0 / framerate for each
>> line and produced the sum. Easy no?
>>
>> My straightforward Haskell solution was:
>> -------------------------------------------------------------------
>> import Text.Regex.Posix
>>
>> main = do
>> f_lines <- readFile "test.out" >>= return . lines
>> let duration = foldl add_line 0.0 f_lines
>> add_line sum line =
>> let [[_,m]] = line =~ "([0-9.]+)"
>> framerate = read m
>> delta = 1.0 / framerate
>> in sum + delta
>> putStrLn $ "Duration (sec): " ++ show duration
>
> Could you try using lazy and/or strict bytestrings. For *any* IO
> performance issue, that should be your first stop.
>
> See, e.g., this sum-file example,
>
> http://shootout.alioth.debian.org/gp4/benchmark.php?test=sumcol&lang=ghc&id=6
>
> which outperforms C on the benchmark.
>
> The second point would be to avoid Text.Regex.Posix, which is
> algorithmically much worse than the PCRE regexes used by perl. Instead,
> try the pcre-light or regex-pcre packages, which have both better
> complexity, and operate on efficient bytestrings.
>
> Finally, there's some small details about writing efficient loops. Give
> type declarations for atomic types like Int and Double, and use strict
> folds or explicit recursion, as in this post,
>
> http://cgi.cse.unsw.edu.au/~dons/blog/2008/05/16#fast
>
> For your task here, with those simple changes, it should be relatively
> easy to produce competitive-with-C performance.
>
> -- Don
>

--
-Corey O'Connor
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

Re: More fun with micro-benchmarks and optimizations. (GHC vs Perl)

2008-07-23T11:18:49Z

coreyoconnor:

> I have the need to regularly write tiny programs that analyze output
> logs. The output logs don't have a consistent formatting so I
> typically choose Perl for these tasks.
>
> The latest instance of having to write such a program was simple
> enough I figured I'd try my hand at using Haskell instead. The short
> story is that I could quickly write a Haskell program that achieved
> the goal. Yay! However, the performance was ~8x slower than a
> comparable Perl implementation. With a lot of effort I got the Haskell
> version to only 2x slower. A lot of the optimization was done with
> guesses that the performance difference was due to how each line was
> being read from the file. I couldn't determine much using GHC's
> profiler.
>
> I still have two questions after all this:
> - Can I get a Haskell implementation as fast as the Perl?
> - What do I need to do to get GHC's profiler to provide me usable
> information? Telling me that 98% of the time was in "main" is not very
> enlightening ;-)
>
> All the code and data for this little experiment I've placed here:
> http://tothepowerofdisco.com/repo/sum_optimization/
> My first, and shortest, version is SumTiny.hs. The optimized version
> is SumFast.hs.
>
> The long version for the curious:
>
> The (cleaned up) data was a 78k line file consisting of lines like the
> following:
> framerate (prev == no pts): 15
> framerate (delta): 25
> framerate (invalid timebase): 12.5
> ... and so on.
>
> The need was for a program that calculated 1.0 / framerate for each
> line and produced the sum. Easy no?
>
> My straightforward Haskell solution was:
> -------------------------------------------------------------------
> import Text.Regex.Posix
>
> main = do
> f_lines <- readFile "test.out" >>= return . lines
> let duration = foldl add_line 0.0 f_lines
> add_line sum line =
> let [[_,m]] = line =~ "([0-9.]+)"
> framerate = read m
> delta = 1.0 / framerate
> in sum + delta
> putStrLn $ "Duration (sec): " ++ show duration

Could you try using lazy and/or strict bytestrings. For *any* IO
performance issue, that should be your first stop.

See, e.g., this sum-file example,

http://shootout.alioth.debian.org/gp4/benchmark.php?test=sumcol&lang=ghc&id=6

which outperforms C on the benchmark.

The second point would be to avoid Text.Regex.Posix, which is
algorithmically much worse than the PCRE regexes used by perl. Instead,
try the pcre-light or regex-pcre packages, which have both better
complexity, and operate on efficient bytestrings.

Finally, there's some small details about writing efficient loops. Give
type declarations for atomic types like Int and Double, and use strict
folds or explicit recursion, as in this post,

http://cgi.cse.unsw.edu.au/~dons/blog/2008/05/16#fast

For your task here, with those simple changes, it should be relatively
easy to produce competitive-with-C performance.

-- Don
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe

More fun with micro-benchmarks and optimizations. (GHC vs Perl)

2008-07-23T11:10:05Z

I have the need to regularly write tiny programs that analyze output
logs. The output logs don't have a consistent formatting so I
typically choose Perl for these tasks.

The latest instance of having to write such a program was simple
enough I figured I'd try my hand at using Haskell instead. The short
story is that I could quickly write a Haskell program that achieved
the goal. Yay! However, the performance was ~8x slower than a
comparable Perl implementation. With a lot of effort I got the Haskell
version to only 2x slower. A lot of the optimization was done with
guesses that the performance difference was due to how each line was
being read from the file. I couldn't determine much using GHC's
profiler.

I still have two questions after all this:
- Can I get a Haskell implementation as fast as the Perl?
- What do I need to do to get GHC's profiler to provide me usable
information? Telling me that 98% of the time was in "main" is not very
enlightening ;-)

All the code and data for this little experiment I've placed here:
http://tothepowerofdisco.com/repo/sum_optimization/
My first, and shortest, version is SumTiny.hs. The optimized version
is SumFast.hs.

The long version for the curious:

The (cleaned up) data was a 78k line file consisting of lines like the
following:
framerate (prev == no pts): 15
framerate (delta): 25
framerate (invalid timebase): 12.5
... and so on.

The need was for a program that calculated 1.0 / framerate for each
line and produced the sum. Easy no?

My straightforward Haskell solution was:
-------------------------------------------------------------------
import Text.Regex.Posix

main = do
f_lines <- readFile "test.out" >>= return . lines
let duration = foldl add_line 0.0 f_lines
add_line sum line =
let [[_,m]] = line =~ "([0-9.]+)"
framerate = read m
delta = 1.0 / framerate
in sum + delta
putStrLn $ "Duration (sec): " ++ show duration
-------------------------------------------------------------------

Just for fun I decided to compare it to a Perl implementation:
-------------------------------------------------------------------
#!/usr/bin/perl
my @f_lines = split(/\n/,`cat test.out`);
my $sum = 0.0;
foreach(@f_lines)
{
/([0-9.]+)/;
my $delta = 1.0 / $1;
$sum += $delta;
}
print("Duration (sec): ", $sum, "\n");
-------------------------------------------------------------------

(I'm sure there are other ways to write this same program in both languages.)
I was pretty happy with how the Haskell implementation's code compared
to the Perl implementation's just in terms of looks. Though I think
the Perl implementation is easier to understand at the part where the
regex match is extracted.

For fun I compared the performance of the two:
(64bit linux running on a 2.4ghz Core2Duo)

$ time perl ./Sum.pl
Duration (sec): 3155.62666666438

real 0m0.121s
user 0m0.103s
sys 0m0.016s

$ time ./SumTiny
Duration (sec): 3155.626666664377

real 0m1.099s
user 0m1.073s
sys 0m0.009s

Youch! ~1s is fast enough that I don't care, but I'm still curious why
there is a 8x performance difference. Profiling with manual cost
center annotations (See SumTinyProf.hs) indicated the "do" expression
main is equal to was responsible 90% of the tiny. Which isn't
revealing at all!

Some experimenting led me to find a 2x slower implementation
(SumFast.hs). Quick notes ont he changes I made:
- Used ByteStrings instead of String.
- Used hGetLine instead of reading the entire file lazily and splitting on "\n"
- Fusing the foldl' with the loop of reading each line from the file.
- Using bang patterns to make the fused loop strict on the
accumulator argument.

I think the largest performance gain was from changing how each line
was read from the file. The lazy read and split seemed very slow
compared to a loop that used hGetLine.

--
-Corey O'Connor
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@...
http://www.haskell.org/mailman/listinfo/haskell-cafe