ByteString gains most improvements as String must be converted o CString

first, internaly, in regex (this is warpper for libpcre), while ByteString not.

libpcre is much faster than posix (I guess posix is also wrapper).

Interface for libpcre is same as for Posix, there is no real effort

in replacing it.

> Date: Tue, 12 Feb 2013 20:32:01 -0800
> From: briand@aracnet.com
> To: nicolasbock@gmail.com
> CC: bmaxa@hotmail.com; bob@redivi.com; haskell-cafe@haskell.org
> Subject: Re: [Haskell-cafe] performance question
>
> On Tue, 12 Feb 2013 15:57:37 -0700
> Nicolas Bock <nicolasbock@gmail.com> wrote:
>
> > > Here is haskell version that is faster than python, almost as fast as c++.
> > > You need to install bytestring-lexing package for readDouble.
>
>
> I was hoping Branimir could comment on how the improvements were allocated.
>
> how much is due to text.regex.pcre (which looks to be a wrapper to libpcre) ?
>
> how much can be attributed to using data.bytestring ?
>
> you have to admit, it's amazing how well a byte-compiled, _dynamically typed_ interpreter can do against an actualy native code compiler. Can't regex be done effectively in haskell ? Is it something that can't be done, or is it just such minimal effort to link to pcre that it's not worth the trouble ?
>
>
> Brian
>