Comments on: What the heck is a Wide Finder, anyway?

By: the drive-by reader

the drive-by reader — Sun, 03 Feb 2008 10:59:11 +0000

Nice speedup.

It would be interesting to see how Haskell compares to a similarly sped-up python version.

By: Alexy

Alexy — Mon, 12 Nov 2007 09:35:36 +0000

Bryan — many thanks for very instructive Haskell. Is there a complete version to download and benchmark with both GHC 6.6.1 and 6.8.1?

By: Bryan O'Sullivan

Bryan O'Sullivan — Thu, 18 Oct 2007 06:39:45 +0000

Oops, yes, typo there. And I’m using a newer version of the regexp libraries, which do work with lazy ByteStrings.

By: Greg

Greg — Thu, 18 Oct 2007 06:33:15 +0000

I think these two statements are true:

1. ghc 6.6 knows about about Text.Regex.Posix, but not about Text.Regex.POSIX.

2. Text.Regex.Posix doesn’t know about lazy bytestrings:

“No instance for (Text.Regex.Base.RegexLike.RegexLike Text.Regex.Posix.Wrap.Regexe
Data.ByteString.Base.LazyByteString”

Or am I misunderstanding something? The example compiles with Data.ByteString.Char8

Thanks again.

By: Bryan O'Sullivan

Bryan O'Sullivan — Tue, 16 Oct 2007 04:01:08 +0000

Greg, here’s the list of imports.

import Control.Monad (forM_)
import qualified Data.ByteString.Lazy.Char8 as LB
import Data.List (foldl’, sortBy)
import qualified Data.Map as M
import System.Environment (getArgs)
import Text.Regex.POSIX ((=~))

The result of (=~) is a list of lists, as the first pattern in the case expression indicates. Each element of the inner list is a lazy bytestring, representing a match subgroup.

By: Greg

Greg — Tue, 16 Oct 2007 00:45:52 +0000

Can you give some hints about how to get the two-minute version to compile?

Most of the imports are easily guessed except LB. LB should include line, readFile, and ByteString, but these seem to be in Prelude, Prelude, and Data.ByteString.

Looking ahead, how is the polymorphism for =~ resolved?

By: Kig

Kig — Tue, 02 Oct 2007 16:44:35 +0000

Really cool post.
Things like this make me happy that you’re part of the “Real World Haskell” thing, because this is exactly the kind of code that reminds me that Haskell is useful for day-to-day code-duggery.

By: Itkovian

Itkovian — Tue, 02 Oct 2007 11:40:27 +0000

Not to nitpick, but 7.7 is hardly twice as fast as 11.6 (Python). Nice job though.

By: Luke Plant

Luke Plant — Fri, 28 Sep 2007 00:27:23 +0000

A nit: isn’t “compare `on` snd” the same as “comparing snd” (using Data.Ord)? You can remove the definition of ‘on’ and save a line.