Re: remove DnD image before drop

[Enrico Sersale]

On Wed, 23 Jan 2002, Adam Fedor wrote:

> Enrico Sersale wrote:
>
> >
> > And the crash confirming from keyboard on the alert panel remains...
> >
> > #0  0x403d8c20 in _NSRemoveHandler (handler=0xbfffe630) at
> > NSException.m:272
> > #1  0x40412a1f in -[NSRunLoop acceptInputForMode:beforeDate:]
> > (self=0x81b9620,
> >     _cmd=0x404c0880, mode=0x819bb40, limit_date=0x85f7550) at
> > NSRunLoop.m:1424
>
> This is a very unusual place for a crash to occur. The only way this
> could happen is if the default (or current) thread has not been properly
> initialized, which means there is something very wrong with the base
> library.

I'm not an expert of the Libraries internals, but, don't the next lines
leave also the possibility of a bug in xgps?
The crash happens on a key-down event; if you confirm with the mouse, all
goes well...

#3  0x40126299 in -[NSGraphicsContext(Ops)
DPSGetEventMatchingMask:beforeDate:inMode:dequeue:] (self=0x824e228,
_cmd=0x407b6f50, mask=4294967295,
    limit=0x81ba9e8, mode=0x819bb20, flag=1 '\001') at
NSGraphicsContext.m:2364
#4  0x407753a0 in -[XGContext(X11Ops)
DPSGetEventMatchingMask:beforeDate:inMode:dequeue:] (self=0x824e228,
_cmd=0x4027ed80, mask=4294967295, limit=0x81ba9e8,
    mode=0x819bb20, flag=1 '\001') at SharedX/XGContextEvent.m:1729
#5  0x400c12d6 in DPSGetEvent (ctxt=0x824e228, mask=4294967295,
    limit=0x81ba9e8, mode=0x819bb20) at
../Headers/AppKit/DPSOperators.h:2563


> It's more likely that you failed to clean/recompile a library (base,
> gui, your own) or the application after a large change, such as the
> addition of an ivar.

No, after some problems I've had some month ago, every time I update from
CVS, I recompile all, make, base, gui and xgps.
And GWorkspace is rebuild from a cleaned source some times/days.

> I tried the same procedure and got a crash also (with the 0.2.7
> sources), but for a different reason which the attach patch partial
> fixes. Unfortuneatly, there is also another problem with a Watcher
> object being released too soon that causes it to crash again soon after.
> I haven't had time to track this down.

Thank you for the patch. That error was in many other sources, too. This
because often I copy parts of code in an other file...

Regarding the Watcher, how have you discovered this bug? How can I
reproduce the situation?