Date: Tue, 4 Jan 2000 15:53:45 +0200 (WET)
From: Andris Pavenis <pavenis AT lanet DOT lv>
To: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
cc: Nicola Bortolotti <nbsoft AT libero DOT it>, djgpp AT delorie DOT com
Subject: Re: Commport problems in *two* Win95 dosboxes
In-Reply-To: <Pine.SUN.3.91.1000104110433.26733G-100000@is>
Message-ID: <Pine.A41.4.05.10001041549120.45538-100000@ieva01.lanet.lv>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Reply-To: djgpp AT delorie DOT com
X-Mailing-List: djgpp AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk



On Tue, 4 Jan 2000, Eli Zaretskii wrote:

> 
> On Mon, 3 Jan 2000, Nicola Bortolotti wrote:
> 
> > But with two DJGPP apps running concurrently, the performances become very
> > worse than Clipper (a P-Code pseudo-compiled language with a freeware add-on
> > to drive commports!). The underruns make the PCM files pretty unaudible,
> > severely crackling!
> > 
> > What a surprise! The revenge of the runtime interpreted Clipper code...
> > 
> > It seems like DJGPP apps don't like Windows DosBoxes in background (in
> > foreground they run fine but one at a time). Including __dpmi_yield() in the
> > outportb cycle just hangs the execution...
> > 
> > Please note that the commport output is unbuffered in both DJGPP and Clipper
> > cases. Under DJGPP I've used SVAsync library.
> 
> I cannot possibly claim to understand all the intricacies of your
> non-trivial application, but I thought you need to know the following
> gotchas related to Windows 9X.
> 
> First, Windows by default suspends a DOS box which goes into
> background.  To avoid this, you need to uncheck the "Background:
> Always suspend" box in the DOS box's property sheets (look under the
> "Misc" tab).
> 
> Second, amazingly enough, a DOS box in the background doesn't get
> timer interrupts delivered to it.  I don't know if this happens with
> other hardware interrupts as well, like the async port interrupt (I
> assume you use it), but it might be a good idea to check this.

My experience shows that it gets part of timer interrupts in background
(my rough estimate is 20% or less depending on system load). I'm getting
such results in one my DOS real mode TSR program which hooks timer
interrupt (or different interrupt when corresponding hardware is
available, but I want to be able to develop programs without access to
real hardware...)
 

> 
> Finally, it might be some bug in SVAsync.  If nothing else helps,
> maybe you should consider to try an alternative library.  Section 22.3
> of the DJGPP FAQ lists several such libraries.
>