From: horst DOT kraemer AT gmx DOT de (Horst Kraemer)
Newsgroups: comp.os.msdos.djgpp
Subject: Re: qsort's algorithm
Date: Mon, 23 Oct 2000 01:13:12 GMT
Lines: 19
Message-ID: <39f35c64.3093928@news.cis.dfn.de>
References: <tf4qusk3quottc52jj9cmuhmqlvk48cgm7 AT 4ax DOT com>
NNTP-Posting-Host: a0f6e.pppool.de (213.6.15.110)
X-Trace: fu-berlin.de 972263549 22565037 213.6.15.110 (16 [27606])
X-Newsreader: Forte Free Agent 1.11/32.235
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com

On Wed, 18 Oct 2000 03:02:12 GMT, Damian Yerrick
<Bullcr_pd_yerrick AT hotmail DOT comRemoveBullcr_p> wrote:

> From what I understand of the C standard, qsort() can use any decent
> sorting algorithm.  Does DJGPP libc's qsort() have bad performance on
> already sorted data?

According to the C standard an implementation may choose *any*
algorithm.

DJGPP's qsort doesn't perform worse with already (almost) sorted data.
According to my experience it has a big performance problem if the
number of keys to be sorted is small w.r.t. to the number of items.

Try to sort a big integer array containing only 0 and 1 entries. 

Regards
Horst