Dive list

Dirk Hohndel dirk at hohndel.org
Tue Oct 8 13:28:24 UTC 2013


On Tue, 2013-10-08 at 06:22 -0700, Dirk Hohndel wrote:
> On Tue, 2013-10-08 at 09:40 +0200, Robert Helling wrote:
> > Hi,
> > 
> > sorry I was not available on IRC last night. Latest master does compile for me out of the box and also gets the profile right (no integer meters).
> > 
> > However, I noticed some oddities with the dive list: 
> > 
> > It doesn't get the date right:
> > 
> > https://www.dropbox.com/s/lxqqmsck9xqu90c/Screenshot%202013-10-08%2009.10.38.png
> 
> Oops. That's not the build system. I must have broken that when ripping
> out glib/gtk pieces.
> Kind of funny that I ran Subsurface dozens of time tracking down other
> stuff and never noticed.
> 
> > and when I move around with the mouse and fold/unfold some trips, I can
> > even get to a state where its seems to show format strings directly
> > (that goes away after about a second and goes back to the state above):
> > 
> > https://www.dropbox.com/s/26edrnfje24i3xk/Screenshot%202013-10-08%2009.09.36.png
> > 
> > Unfortunately, this behavior is not really reproducible.
> 
> I'll need to bisect this and figure out how exactly I managed that. I'm
> thinking that it's the translation subsystem changes, but I'm not quite
> sure.

Spot on. I returned a pointed to the raw data inside a QString - but the
QString's scope ended and the memory was freed before things got used...

Oops.

The easy fix (pushed) is to simply strdup() this... but that of course
creates a memory leak with every translation. Not good.

/D



More information about the subsurface mailing list