Edit dive info, cancel, crash

[Dirk Hohndel]

On Sun, Jul 06, 2014 at 05:10:28PM +0200, Henrik Brautaset Aronsen wrote:
> Robert C. Helling wrote:
> >when I open subsurface, get into edit mode (e.g. type a number in the so far empty water temperature field of a dive) and then cancel the edit, subsurface crashes.
> >
> >I have now spent quite some time with the debugger but could not find the origin of the problem. Sometimes, I get a dive where the first event has the “deleted” flag set and which has a next field leading nowhere, in another case the loop in  populate_plot_entries runs forever with really really big “time” variable.
> >
> >I don’t really know how to investigate this further so this is my report.
> 
> 
> I get something similar:
> 
> 1) Start subsurface without any dives
> 2) Add dive and save it
> 3) Enter values in the water temp field to get it in edit mode
> 4) Press cancel & don't save
> 5) Try to enter values in the water temp field again: Subsurface hangs while
> eating 90% CPU

Valgrind shows me an issue with this sequence.
Specifically, we try to copy the samples of our dive as we start to edit
it (loadFromDive()) and then find out that those samples had been freed
the last time we started editing that dive.

I'm trying to track this down...

/D