Re: [lmi] System crashes with invalid input using input sequence GUI

[Vadim Zeitlin]

On Thu, 21 May 2015 22:36:25 +0000 "Boutin, Wendy" <address@hidden> wrote:

BW> I read that documentation, but I'm not sure I understand the unusual 
BW> behavior in the UI. It seems to work normal to me using your patch. Is 
BW> it the case that I'd have to upgrade wxWidgets to see the behavior? If
BW> so, then I will probably avoid that for now and focus only on the lmi 
BW> patch. 

 Yes, sorry for being unclear. The unusual behaviour can only be seen with
fixed wxWidgets version (because otherwise it would crash before showing
it) and without the lmi patch I posted (which avoids putting the UI in this
state entirely). FWIW in lmi case it would have just resulted in a spin
control initially containing 0 but which could be change to contain 1 by
clicking its _down_ arrow -- and clicking the _up_ one would then switch to
0 again. This is clearly intentional, under MSW, but seems very strange to
me.

BW> >  The patch below just disables the spin control if there is no valid range
BW> > for it. I am not sure if it's the best solution UI-wise, but I can't think
BW> > of anything better right now and it does fix the crash (even without the
BW> > updated wxWidgets). 
BW> 
BW> The crash is indeed fixed with that patch and I think disabling the
BW> spin control is the best option. I couldn't come up with a more natural 
BW> solution from a user perspective. 

 Greg, if you're still reading this, I'd like to make a last plea for
applying the original patch. If Wendy confirms that it doesn't make the UI
behave unacceptably, it's definitely an improvement compared to the current
situation. If you'd prefer, I could make an even simpler patch that would
skip disabling the control but just check that it never sets it to a range
with "min > max". Such patch would be 100% safe and I really think it
should be applied to fix the crash.

 Regards,
VZ