Testing beta 3

[Dirk Hohndel]

On Mon, Oct 05, 2015 at 09:50:40PM +0100, Dirk Hohndel wrote:
> On Mon, Oct 05, 2015 at 02:45:04PM +0300, Lubomir I. Ivanov wrote:
> > On 5 October 2015 at 14:30, Steve <stevewilliams at internode.on.net> wrote:
> > >
> > > And the same message box that states:
> > > subsurface.exe has stopped working
> > >
> > 
> > could you please:
> > 
> > press the "Windows key" + R:
> > type: eventvwr
> > 
> > goto WindowsLogs -> Applications.
> > 
> > there should be an Error about Subsurface crashing recently.
> > 
> > copy / paste the text and send it here to the mailing list.
> 
> Faulting application name: subsurface.exe, version: 0.0.0.0, time stamp:
> 0x56114c6e
> Faulting module name: msvcrt.dll, version: 7.0.10240.16384, time stamp:
> 0x559f3e0e
> Exception code: 0xc0000005
> Fault offset: 0x00088fec
> Faulting process id: 0xe18
> Faulting application start time: 0x01d0ffadd569a93f
> Faulting application path: C:\Program Files
> (x86)\Subsurface\subsurface.exe
> Faulting module path: C:\Windows\SYSTEM32\msvcrt.dll
> Report Id: 44baed00-550b-4712-ace1-00da3991e4c4
> Faulting package full name: 
> Faulting package-relative application ID: 
> 
> - why is the version 0.0.0.0 ?
> - what information does this give me?
> 
> That's so entirely not helpful, Microsoft...
> 
> Running it in the debugger shows me that get_local_dir() gets called with
> an invalid remote pointer and passes an invalid pointer to strlen().
> 
> So the bug is in print_files() which calls is_git_repository() with
> dry_run == true which causes that to return dummy_git_repository back to
> us which is defined as 3ul which we definitely should not be
> dereferencing.

And we aren't - my bad. Instead we are dereferencing the remote variable
which never gets initialized when dry_run is set.

I have a fix for this. I'll address a few other things, push this all out
and generate new daily builds...

Yet another bug squashed BEFORE we went to release.

Hooray for people testing the beta!!!!!

/D