Date: Thu, 9 Jul 1998 11:05:01 +0300 (IDT)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To: Mariano Alvarez Fernández <malfer AT teleline DOT es>
cc: Nate Eldredge <nate AT cartsys DOT com>, djgpp AT delorie DOT com
Subject: Re: About DJGPP v2.02
In-Reply-To: <35A3B749.50BC@teleline.es>
Message-ID: <Pine.SUN.3.91.980709110440.13185F-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Content-Transfer-Encoding: 8bit
Precedence: bulk


On Wed, 8 Jul 1998, Mariano Alvarez Fernández wrote:

> Eli Zaretskii wrote:
> > One thing that v2.02 changes is the way buffered file I/O works.
> > v2.02 adapts itself better to the specific use of buffered I/O
> > functions (sequential vs. random access), which should show in
> > programs that seek into files a lot.  Perhaps your program is one of
> > those?
> 
> Yes I seek a lot, the data-bases are shared in a LAN, I can't pre-read
> because any body can update it, I use lock files as semaphores.
> Obviously the new algorithm works great for me :-)

Seeking a lot over a network was precisely the case where the improved
buffering should give you the greatest boost.  I'd expect at least
twofold performance gain in these cases, with the v2.02 library.