[openbox] [nikitin at vnukovo.net: Openbox bug]

[Ava Arachne Jarvis]

[Ben Jansens - Mon,  4 Nov 2002 03:23:27 PM CST]
> Is anyone else able to decipher what this guy is having go wrong..?
> Mozilla is crashing, what?
> 
> Ben
> 
> ----- Forwarded message from Alexey Nikitin <nikitin at vnukovo.net> -----
<snip> 
> File Window.cc, lines 2034, 2035 (you restore premax window geometry):
> 
> blackbox_attrib.premax_w = premax_w;
> blackbox_attrib.premax_h = premax_h;
> 
> After I start Mozilla, it runs with net_wm_state_maximized atoms, and Openbox makes its window
> maximized. It seems to me, premax_x, premax_y, premax_w and premax_h have zero values in 
> BlackboxWindow::remaximize method when it's called from a BlackboxWindow constructor -- 
> so up-clicking "maximize" button can't restore window geometry and Mozilla crashes.

So far as I can determine, 

  - Mozilla, when first started, runs with a maximized state.  Possibly 
    a setting in Mozilla to turn this on...
  - So, Openbox makes the window maximized.  Well, good, and
    NETWM-compliant.
  - But since this window never had any pre-maximized state in Openbox,
    the pre-maximized ints are set to 0.  
  - This leads to, when the user attempts to unmaximize Mozilla,
    Openbox probably resizing the Mozilla window to 0 by 0 pixels, which 
    Mozilla really really doesn't like, and so Mozilla goes boom.

> I've made a dirty fix. You'll make something more elegant, though:
> 
> blackbox_attrib.premax_w = premax_w ? premax_w : 780;
> blackbox_attrib.premax_h = premax_h ? premax_h : 580;

His fix just sets arbitrary starting values for the pre-maximized state
if a window starts out maximized in Openbox...


-- 
| The algorithm for finding the longest path in a graph is NP-complete.
| For you systems people, that means it's *real slow*.
| 		-- Bart Miller