[openbox] Multiple screens not considered correctly in Openbox-3.0-rc3

[Ben Jansens]

On Fri, Oct 10, 2003 at 04:15:15AM -0400, Kacper Wysocki wrote:
> >> The new feature in rc3, the black frame border while window cycling   
> >> was a very nice idea indeed.
> 
> >It's really stolen from metacity :) I can't take credit for it.
> 
> It's still very nice, and I know you implemented and released it within  
> three days of hearing of it. Or is it a pure cut-n-paste job? ;-)

Err, no, our code is not similar at all.

> >Ya, use xinerama, there is no support for dual head in non-xinerama  
> >setups. Changing this would be extremely non-trivial at this point so  
> >its going to stay that way. However, take heart, Openbox is pretty  
> >smart about xinerama, and ideas for further smartness would be  
> >appreciated.
> [snip]
> >Running 2 Openbox's does not work. They end up fighting over focus  
> >since they both keep losing track of the focus (since it disappears  
> >off to the other screen).
> >
> >Sorry, you'll have to learn to love xinerama or run Openbox 2 or
> >something.
> 
> Okay, so I'm "learning to love" xinerama, for now.
> 
> I'm sorry to hear that there will be no non-xinerama openbox 3. Running  
> openbox 2 again is impractical, since I now have some applications (for  
> example gDeskCal) that don't work with ROX' compatability mode, and old  
> openbox doesn't work properly without rox' compatability mode.
> Um, I assume a developer for a xinerama-supporting window manager uses  
> xinerama on a daily basis for his work?

Nope, I would if I had any money with which to throw at a second monitor. I
can borrow a monitor for testing purposes however, and did to write it all.

> The biggest trouble with xinerama isn't that window managers aren't  
> trying to be smart about it, because most are, but that X applications  
> are super-dumb about xinerama. In a common xinerama setup where the two  
> screens aren't the same model, one screen will invariably run at a  
> higher resolution than the other, resulting in rox, and other  
> applications attached to the bottom of the screen to disappear into the  
> deadzone created by the difference in resolutions. I haven't seen any  
> windowmanager handle this properly, which makes me think maybe this is  
> a xinerama limitation.

Hm Im not sure what I should do about that. Filling only one head with the
desktop doesn't seem a right approach.

> A second problem is that some applications adjust their initial width/ 
> height according to the desktop area, which makes for some pretty  
> ridiculously x-stretched windows. Maximising a window over two screen  
> is fun, but confusing and perfectly useless.

Well, Openbox maximizes on one head.

> 
> Lastly, sort of on a side note, openbox has some issues with certain  
> applications maximising/restoring in xinerama. At least 4 out of 5  
> times openbox will crash if I fullscreen, then restore, then fullscreen  
> again a movie in xine, colorfully stating:
> "Fuck yah. Core dump. (Signal=11)"
> Where did this core go, anyways? (Ideally, though, xine would play the  

http://openbox.org/bugs.php

> movie in duplicate on both screens at once. Irrelevant, since I'm  
> pretty sure this is driver-depenent and has to do with the way video  
> overlay works.)

That shouldn't crash Openbox.

> Sometimes a window will maximise to a different screen than it's  
> running on. I guess it must be a focus thing, but I'm not quite sure.  
> It's hard to consistently reproduce these bugs.

It is probly a few pixels over onto the primary or soemthing. That code
could use more intelligence.

> I'll certainly try working with xinerama and keep posting any  
> difficulties I have here, if it helps.

Sure :)

To be perfectly honest, I'd rather support 2 screens than xinerama, but
that's just not practical for me at this time. It would pretty much require
a rewrite/organization of the entire codebase.

Ben
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://icculus.org/pipermail/openbox/attachments/20031010/3f373fd4/attachment.pgp>