🚀 go-pugleaf

While debugging a MSC 7.0 (pre Win32) DOS program with CodeView 4.00 
running in a DOS box under Windows 3.1 enhanced mode with the 2-monitor 
option set I get the following behavior if I exit CV immediately after 
executing the MS extention _setvideomode():

_setvideomode() argument        behavior

_HRES16COLOR			Mouse works, but mouse cursor is 
_VRES16COLOR                    invisible in PWB.  Shelling to DOS 
                                and resetting the video mode to 3 
                                doesn't fix it.

_XRES16COLOR			Next execution of CV reverses the
				functions of the two monitors:
                                CV comes up on SVGA and C program
                                tries to run on mono monitor.

I haven't taken the latter case all the way to the _setvideomode()
function for fear it would damage my mono, but it's good
I was paying attention.  The hardware is a 486/33 clone with 
a ViewSonic 5+ monitor and Diamond SpeedStar SVGA with 1 MB and a
Treu Bleu mono adapter and monitor and MS mouse. 

The situation when running PWB from MSDOS 5.0 is worse:  As soon as
CV exits back to PWB, PWB reports an "Internal VM error PWB 3912" and
exits to DOS (probably reports this under Win also, but the video flips
so fast I can't read it).  Reloading PWB at this point still shows the 
invisible mouse cursor error (I didn't try _XRES, all this recompiling 
is a pain).  Only  ^<Alt-Del> gets the cursor back to normal.  

One solution is never to exit CV without allowing the C program to 
execute the call to _setvideomode(_DEFAULTMODE) often on cleanup.
Of course, this works only if the program makes it all the way to that
instruction.  Even if the program works well enough to exit normally 
(mine sometimes do :)) I'm often in such a hurry to fix a found
bug that I exit CV ASAP instead of letting the program exit normally 
first.  Frustration ensues.

Why can't CV read the state of the adapters on entry and restore them 
on exit just in case the program doesn't make it all the way to the 
cleanup (rhetorical question ;-( )!  What causes this behavior?  
Why is it different for _XRES?  Any ideas how to correct the problems 
inside PWB without exiting and reloading the program (Win) or doing the
boyscoutsalute (DOS)?  I have the Win32/NT beta but don't have a CDROM
yet so it's still on the disk.  Is this fixed in that version?

-- 
Doug Fitts
dfitts@u.washington.edu
First in Thirst