[Haskell-cafe] Differences in optimisiation with interactive and compiled mo

[Daniel Fischer]

Am Samstag, 10. Dezember 2005 18:29 schrieb Branimir Maksimovic:
> From: Tomasz Zielonka <tomasz.zielonka at gmail.com>
>
> >To: Branimir Maksimovic <bmaxa at hotmail.com>
> >CC: lemming at henning-thielemann.de, haskell-cafe at haskell.org
> >Subject: Re: [Haskell-cafe] Differences in optimisiation with interactive
> >and compiled mo
> >Date: Sat, 10 Dec 2005 18:14:58 +0100
> >
> >On Sat, Dec 10, 2005 at 04:14:20PM +0000, Branimir Maksimovic wrote:
> > > Nice code.
> >
> >But incorrect. I have broken it when refactoring :-/
> >
> >Here is the correct version:
> >
> >replace2 src dst = repl
> >   where
> >     repl input | src `isPrefixOf` input = dst ++ repl (drop (length src)
> >input)
> >     repl (x:xs) = x : repl xs
> >     repl [] = []
> >
> > > But it takes lot of ram (1GB is not enough )and can't execute my test.
> >
> >Can you check this version?
>
> It's ok now; 2 megs like other versions.
> It's just about 1.5 seconds slower then mine version and Daniels version is
> a
> bit faster the mine.
>
> bmaxa at MAXA ~/tutorial
> $ time ./replace1.exe
> Working:seaseasearch replace  able seaseaseasearch baker seaseasearch
> charlie
> True
> Done
>
>
> real    0m14.140s
> user    0m0.015s
> sys     0m0.000s
>
> Greetings, Bane.
>
On my thingy, Tomasz' version is a bit faster than my version of the same 
algorithm for seasea..., and a bit slower for rrrrrrrrrrrrrrrr... and this 
algorithm is definitely the fastest submitted. Odd that your timings are 
different (in order) -- maybe it's something about Linux vs. Windows?

Cheers,
Daniel