Date: Tue, 7 Mar 1995 07:16:34 +0500
From: ld AT netrix DOT com
To: djgpp AT sun DOT soe DOT clarkson DOT edu, A DOT D DOT Brown AT bradford DOT ac DOT uk
Subject: Re: Mice, interrupts & dos 5


> From djgpp-bounces AT sun DOT soe DOT clarkson DOT edu Mon Mar  6 15:24 EST 1995
> From: A DOT D DOT Brown AT bradford DOT ac DOT uk
> Subject: Mice, interrupts & dos 5
> To: djgpp AT sun DOT soe DOT clarkson DOT edu
> Date: Mon, 6 Mar 1995 19:17:20 +0000 (GMT)
> Mime-Version: 1.0
> Content-Transfer-Encoding: 7bit
> 
> 1 - Installing custom mouse handlers
> ------------------------------------
> One of the nice features of the microsoft mouse driver is for it to call
> one of your own routines when an event (e.g. motion or button click) occurs.
> Unfortunatly, it requires a segment:offset form of passing the address of
> your routine, which obviously won't work in protected mode programs.  Is
> there a foolproof method of doing the equivalent?  (I am currently calling
> mouse routines through simulated int 33h, as you would for a 16 bit program.

  You can use go32_dpmi_allocate_callback_retf () to allocate a realmode
callback in conventional memory. Then you can use the int 33h call to make
the mouse call your mouse handler.

> 2 - Interrupts
> --------------
> I know there has been a lot of traffic to do with this recently, but can
> anyone offer practical advice as to how to resolve the problems with real
> and protected mode interrupts.  Just using protected mode interrupts seems
> to work fine on dos 6, but when I tried it using HIMEM & EMM386 with dos 5
> it caused chaos.  (Missing interrupts like nobodys business).  Is there a
> universal method of handling interrupts that will work no matter which
> memory configuration / mode you are working in?

  I'd suggest that you run the program under DPMI (CWSDPMI is out for beta
testings.)