Re: [lmi] Segfault in lmi with wxWidgets-2009-03-10

[Vadim Zeitlin]

On Wed, 11 Mar 2009 06:01:42 +0000 Greg Chicares <address@hidden> wrote:

GC> To reproduce, with luck:

 I definitely don't see this :-(

GC>   start lmi
GC>   open any file of any type, and leave its window open
GC>   close lmi (forcing the open MDI child window to be closed)

 This was one of the scenarios which I checked while testing my patches so
I must have done this at least a dozen times (I had to retest many times
because of the multiple changes) without ever seeing the problem. There
are, of course, many differences between our builds: not only I use MSVC
but I also have some patches applied locally and I'm not sure if you
applied them to the version you use -- did you?

GC> I'm unable to reproduce the problem when I run lmi under gdb. IIRC, gdb
GC> does some zero-initialization that can mask stray-pointer problems.

 I don't know about gdb but when using other Windows debuggers the OS
itself changes the program behaviour in much more important way: if the
program is being debugged, ntdll.dll itself (used by all Windows programs)
detects it and uses the special debug heap instead of the normal one. So
the usual reason for the bugs not being reproducible under debugger under
Windows is that heap allocations work differently -- if your version of gdb
supports it, it'd be worth trying to attach to an already running program
instead as this wouldn't change the heap being used. Could you please try
it?

 Thanks,
VZ