Thread: Re: [GD-Windows] More multimon

Brought to you by: vexxed72

gamedevlists-windows

Re: [GD-Windows] More multimon

From: Rich <leg...@xm...> - 2003-03-31 03:34:10

In article <2003330172253.380519@HOOKDELL220>,
    Brian Hook <bri...@py...>  writes:

> This is also a good point.  I don't mind the multimon hacking=
>  about, 
> but the nVidia "I'm just a big wide monitor as far you're=
>  concerned" 
> thing is kind of annoying because it's probably the single most 
> prevalent multimon config out there.

Make sure you tell nvidia.

NVidia has this annoying habit of trying to "fix things" for the
developer that don't need fixing.  Like you write code that's
incorrect and the driver "fixes" it for you behind your back, leaving
you oblivious to the dormant bug hiding in your code.  This sort of
thing just creates more problems than it solves, IMO, and leads to the
creation of more poorly written programs rather than less.
-- 
"The Direct3D Graphics Pipeline"-- code samples, sample chapter, FAQ:
          <http://www.xmission.com/~legalize/book/>
        izfree: Open source tools for Windows Installer
                <http://izfree.sourceforge.net>

[GD-Windows] Some multimon data

From: Brian H. <bri...@py...> - 2003-03-31 06:15:19

I hacked up a quick multimon information extractor (GDI only, no=
 DX 
stuff) and ran it on some of my multimon systems to see what 
interesting information I could glean (or not).

These are all Win2K systems, I'm not sure what happens on XP or 
Win98, but since the APIs are the same I'm assuming that it won't=
 do 
anything too out of whack.  Win98 is my biggest concern since=
 it's a 
different driver model.

System #1: GF3/AGP + Rage128/PCI + TNT2/PCI

GetSystemMetrics( SM_CMONITORS ) =3D 3
\\.\DISPLAY1
   - "NVIDIA GeForce3", attached, primary
   - \\.\DISPLAY1\Monitor0, "Plug and Play Monitor"
      - attached, 181 video modes
\\.\DISPLAY2
   - "Xpert 128", attached
   - \\.\DISPLAY2\Monitor0, "Default Monitor"
      - attached, 210 video modes
\\.\DISPLAY3
   - NVIDIA RIVA TNT2 Mode 64/Model 64 Pro, attached
   - \\.\DISPLAY3\Monitor0, "Default Monitor"
      - attached, 336 video modes
\\.\DISPLAY5
   - NetMeeting driver

That is a pretty sane system.

System #2a: GF4 Ti4200, single virtual desktop

GetSystemMetrics( SM_CMONITORS ) =3D 1
\\.\DISPLAY1
   - "NVIDIA GeForce4 Ti 4200", attached, primary
   - \\.\DISPLAY1\Monitor0, "Plug and Play Monitor"
      - NOT attached, 217 video modes, 2560x1024
\\.\DISPLAY3
   - "NetMeeting Driver"

Pretty bizarre all the way around.  Only a single monitor, only a=
 
single adapter found, and the monitor on DISPLAY1 is listed as 
"unattached", which is either a bug or clearly something I don't=
 
understand.

Next I configured it the same as above, but turned off one of the=
 
desktops, which _should_ make it seem like a complete single=
 monitor 
system.

System #2b: GF4 Ti4200, second monitor not enabled
GetSystemMetrics( SM_CMONITORS ) =3D 1
\\.\DISPLAY1
   - "NVIDIA GeForce4 Ti 4200", attached, primary
   - \\.\DISPLAY1\Monitor0, "Default Monitor"
   - \\.\DISPLAY1\Monitor1, "Default Monitor"
   - \\.\DISPLAY1\Monitor2, "Plug and Play Monitor"
   - \\.\DISPLAY2\Monitor3, "Plug and Play Monitor"
      - attached, 181 video modes
\\.\DISPLAY2
   - "NVIDIA Dualview", not attached
   - \\.\DISPLAY1\Monitor0, "Default Monitor"
   - \\.\DISPLAY1\Monitor1, "Default Monitor"
   - \\.\DISPLAY1\Monitor2, "Plug and Play Monitor"
      - attached, 275 video modes
   - \\.\DISPLAY2\Monitor3, "Plug and Play Monitor"
\\.\DISPLAY4
   - "NetMeeting Driver"

Well, this is a lot weirder than I was expecting.  The first big=
 
thing is that there are four monitors enumerated.  The second=
 thing 
is that it looks like there are two adapters in the system, when=
 
there's only one.  So long as the code observes "primary only 
adapters and attached monitors", it should be okay.

Finally I turned on the checkbox for "Treat each output as a=
 separate 
video device" (or whatever it's called).

System #2c: GF4 Ti4200, "independent" outputs, dual head
GetSystemMetrics( SM_CMONITORS ) =3D 2
\\.\DISPLAY1
   - "NVIDIA GeForce4 Ti 4200", attached, primary
   - \\.\DISPLAY1\Monitor0, "Default Monitor"
   - \\.\DISPLAY1\Monitor1, "Default Monitor"
   - \\.\DISPLAY1\Monitor2, "Plug and Play Monitor"
   - \\.\DISPLAY2\Monitor3, "Plug and Play Monitor"
      - attached, 181 video modes
\\.\DISPLAY2
   - "NVIDIA Dualview", attached
   - \\.\DISPLAY1\Monitor0, "Default Monitor"
   - \\.\DISPLAY1\Monitor1, "Default Monitor"
   - \\.\DISPLAY1\Monitor2, "Plug and Play Monitor"
      - attached, 275 video modes
   - \\.\DISPLAY2\Monitor3, "Plug and Play Monitor"

Finally, a little bit of sanity -- this works just like #2b,=
 except 
the second display is marked as attached.

So the heuristic that I'll probably use:

if more than one monitor reported
   enumerate monitors
      get primary
         get attached
else
   if aspect ratio is > 2:1
      assume single virtual desktop multimon system
   else
      treat as a regular old single head system

The major thing is that you just have to let the user choose to=
 run 
windowed for the SVD configuration or let them select "Don't=
 change 
video modes" or something similar.  That's just a shitty=
 situation to 
be in unfortunately.

Brian

ATI data was Re: [GD-Windows] Some multimon data

From: Brian H. <bri...@py...> - 2003-04-01 04:00:24

Here's another sample point submitted by someone else for an ATI=
 
Radeon 8500 dual head.

GetSystemMetrics( SM_CMONITORS ) reports 2 monitors

\\.\DISPLAY1, "RADEON 8500 SERIES", attached, primary
   \\.\DISPLAY1\Monitor0, "Plug and Play Monitor"
      - attached
   \\.\DISPLAY1\Monitor1, "Plug and Play Monitor"
      - attached, multidriver
\\.\DISPLAY2, "RADEON 8500 SERIES", attached
   \\.\DISPLAY2\Monitor0, "Plug and Play Monitor"
      - attached, multidriver
   \\.\DISPLAY2\Monitor1, "Plug and Play Monitor"
      - attached

This is fairly alarming, because to a naive multimon inspection=
 it 
looks like there are two adapters, each with two monitors.  Both=
 
adapters are marked as attached, and each has two attached=
 monitors.  
The only difference is that the "multidriver" bit is set in the=
 flags 
field, which NVidia does not set anywhere that I'm aware of.

This isn't a documented bit for the DISPLAY_DEVICE::StateFlags 
structure, so I'm not even sure what it means other than it's=
 related 
to multihead adapters somehow?

Brian

[GD-Windows] Hijacking application

From: Brian H. <ho...@py...> - 2003-08-19 01:51:27

I have a feeling this can't be done, but I figured I'd ask the list 
anyways just in case.

Say I want to write a plug-in for random application XYZ.  XYZ, 
however, may not have a plug-in API or it may be broken.  Use Notepad 
as an example.

If I wanted to add a plug-in, it seems like it would be pretty much 
impossible unless my plug-in didn't require any of the application's 
other data.  

I swore there were a few plug-ins I've seen that have forced their way 
into an application, but I can't think of what they might be now.  The 
obvious ones are like Visual Assist, Max plug-ins, Maya plug-ins, 
Photoshop plug-ins, VS plug-ins, etc. but those all have registered 
APIs.  I've seen screenshots of people that have retrofitted a new 
menu into Notepad just to prove it can be done, but I'm not sure what 
the point of that is.

The typical way you do this is a pretty convoluted process involving 
subclassing and message hooks, but even then you can only trap new 
menu items and maybe a couple of key messages like WM_DESTROY.  You 
obviously can't grab application data and do anything meaningful with 
it.

Brian

Re: [GD-Windows] Hijacking application

From: Kent Q. <ken...@co...> - 2003-08-19 04:17:56

Well, under Windows, if you can get the window handle, there's a lot you 
can hook, and you can inspect most of the standard windows components and 
extract the data they display. There are QA "robot" types of applications 
that can record and play back all sorts of things, and then examine the 
contents of the various components to see if they contain what they should.

But the more an app uses nonstandard controls, the less you can do with it. 
And if there's graphics or internal data that doesn't show up readily in 
the UI, it's pretty tough to imagine how you'd get in by other than 
explicit reverse-engineering.

         Kent


At 09:51 PM 8/18/2003 -0400, you wrote:
>I have a feeling this can't be done, but I figured I'd ask the list
>anyways just in case.
>
>Say I want to write a plug-in for random application XYZ.  XYZ,
>however, may not have a plug-in API or it may be broken.  Use Notepad
>as an example.
>
>If I wanted to add a plug-in, it seems like it would be pretty much
>impossible unless my plug-in didn't require any of the application's
>other data.
>
>I swore there were a few plug-ins I've seen that have forced their way
>into an application, but I can't think of what they might be now.  The
>obvious ones are like Visual Assist, Max plug-ins, Maya plug-ins,
>Photoshop plug-ins, VS plug-ins, etc. but those all have registered
>APIs.  I've seen screenshots of people that have retrofitted a new
>menu into Notepad just to prove it can be done, but I'm not sure what
>the point of that is.
>
>The typical way you do this is a pretty convoluted process involving
>subclassing and message hooks, but even then you can only trap new
>menu items and maybe a couple of key messages like WM_DESTROY.  You
>obviously can't grab application data and do anything meaningful with
>it.
>
>Brian
>
>
>
>
>-------------------------------------------------------
>This SF.Net email sponsored by: Free pre-built ASP.NET sites including
>Data Reports, E-commerce, Portals, and Forums are available now.
>Download today and enter to win an XBOX or Visual Studio .NET.
>http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01
>_______________________________________________
>Gamedevlists-windows mailing list
>Gam...@li...
>https://lists.sourceforge.net/lists/listinfo/gamedevlists-windows
>Archives:
>http://sourceforge.net/mailarchive/forum.php?forum_idU5

Kent Quirk, CTO, CogniToy
ken...@co...
http://www.cognitoy.com