Xref: news2.mv.net comp.os.msdos.djgpp:3119
From: Shawn Hargreaves <slh100 AT york DOT ac DOT uk>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: Allegro Mouse In Win95?
Date: Thu, 25 Apr 1996 16:18:00 +0100
Organization: The University of York, UK
Lines: 49
Message-ID: <Pine.SGI.3.93.960425160602.5424A-100000@tower.york.ac.uk>
References: <Pine DOT SUN DOT 3 DOT 91 DOT 960421113451 DOT 14512Z-100000 AT is>
NNTP-Posting-Host: tower.york.ac.uk
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: <Pine.SUN.3.91.960421113451.14512Z-100000@is>
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp

On Sun, 21 Apr 1996, Eli Zaretskii wrote:

> On Sun, 21 Apr 1996, Orlando Andico wrote:
> 
> > On 20 Apr 1996, David J. Calvin wrote:
> > 
> > > the mouse moves incredibly slowly.  It's fine in Dos, or in Windows
> > > before my program is loaded, but as long as I'm using my program,
> > > the sensitivity is horrible.  This happens with the Allegro
> > > "grabber" program as well, so its
>
> > I'm just guessing here, but I noticed that when I run the JED editor 
> > (compiled with DJGPP) under Win95, everything else slows to a crawl. This 
> > didn't happen under Win3.1, my guess is that the DPMI functionality in 
> > Win95 stinks.
> 
> As long as we are guessing, here's another guess: it might be that all the
> apps that slow Win95 to a crawl just aren't nice DPMI clients in a
> multitasking environment.  If a program polls some resource (like the
> mouse or the keayboard) and doesn't issue `__dpmi_yield' while waiting, it
> might have similar effect on Win95.  (E.g., such problems were reported

a) I don't think much of a 'preemptive' multitasking OS that required
   applications to explicitly yield control to it :-)

b) If I'm right, the problem was with the mouse response _within_ the
   Allegro program, not with the rest of the Windows system (correct me if
   I'm wrong, David). Allegro just uses the int33 DOS mouse driver and
   creates an RMCB for tracking movement and displaying the pointer. It
   then goes off and does other stuff without yielding, but it is a game
   library and most games aren't concerned about the peformance of the
   rest of the OS while they are running... If the mouse is moving slowly
   inside Allegro, my guess is that interrupts are being missed for some
   reason. I have a feeling the djgpp libs create non-reentrant RMCB's,
   but that shouldn't be a problem since it would just cause a little skip
   in the mouse pointer movement, not an actual slowdown. I can't think
   what would be causing interrupts to be completely missed...

> I think it's best to double-check our own applications and libraries 
> before blaming it on products of others (even if it's Microsoft).

I have to agree. Although Microsoft have been responsible for some really
awful code, they spent a lot more $$$ testing Win95 that I did testing
Allegro :-)


Shawn Hargreaves.        Why is 'phonetic' spelt with a ph?
Check out Allegro and FED on http://www.york.ac.uk/~slh100/