Re: cairo drawing problem

[Ivan Vučica]

On Wed, Jun 17, 2020 at 10:09 AM Andreas Höschler via Discussion list
for the GNUstep programming environment <discuss-gnustep@gnu.org>
wrote:
> > Then I saw the rest of the code you attached. My thought goes in the 
> > direction of ‘_sharedMapView’-related code. Even though you’re allocating a 
> > whole new window only when the map view is not shared, you seem to be 
> > releasing window unconditionally.
>
> The ivar window is either nil (in case of _sharedMapView != nil) or 
> initialised with window = [[NSWindow alloc] init...] (in case of 
> _sharedMapView == nil), so the [window release] should be fine!?

This is very nonobvious, and if I were doing code review for this, I'd
recommend code cleanup.

> This stack trace does not look very helpful (at least to me). :-(

Ah, I should have expected this. X's async nature means errors can get
detected 'late'.

Sadly, I have no better thoughts on how to triage which X resource is
getting destroyed.