[wx-dev] Re: Bug in wxtiff library (Vadim Zeitlin)

[Aron Helser]

I have seen this exception reported by Visual Studio 2005 when using 
wxWidgets versions 2.6 and 2.8, and after investigating I determined it 
was harmless - it is a 'first-chance' exception that is reported to the 
debugger, but does not have further consequences for user code. I also 
followed the work on updating libtiff in the trunk, and I believe it 
will fix this minor problem.

My conclusion was that I can ignore this report by Visual Studio until 
the the 3.0 release of wxWidgets, when it should be fixed.

Hope this helps,
Aron

> Date: Sun, 27 Apr 2008 01:09:23 +0200
> From: Vadim Zeitlin <vadim at wxwidgets.org>
> Subject: Re: [wx-dev] Bug in wxtiff library
> To: wx-dev at lists.wxwidgets.org
> Message-ID: <20080426230924.96C9C314574 at riobu.com>
> Content-Type: TEXT/PLAIN; CHARSET=US-ASCII
>
> On Sat, 26 Apr 2008 13:52:56 -0400 Jordan Dauble <jdauble at gmail.com> wrote:
>
> JD> Hello, I am using wxWidgets-2.8.7. When I open tiff files using the wxtiff
> JD> library, several access violations exceptions are displayed in my output
> JD> window in visual studio 2005. The exception occurs in tif_win32.c in the
> JD> _TIFFrealloc function if the p pointer is NULL.
>
>  I don't know whether it's normal for this pointer to be NULL nor whether
> it is a (harmless) exception caught by the GlobalSize() function itself or
> not...
>
> JD> After investigating this
> JD> issue online, I found that someone had posted a solution here
> JD> http://www.asmail.be/msg0055337192.html awhile back. After modifying the
> JD> _TIFFrealloc function as described and recompiling the wxWidgets libraries,
> JD> I no longer receive this exception. Is the wxtiff library not up to date
> JD> with the libtiff library? Should the tif_win32.c file be modified so that
> JD> others do not receive this exception?
>
>  ... but I did backport this to 2.8 branch (it was already ok in the trunk)
> anyhow.
>
>  BTW, I have absolutely no idea what is _TIFFrealloc() doing, the code
> makes absolutely no sense to me and whatever it does it has no bearing to
> the standard realloc() as it always reallocates the memory. As the Unix
> version does use just realloc(), I'm more or less sure that this one is
> completely wrong, if anybody is willing to report this to libtiff
> developers (after checking that it hasn't been corrected yet in the latest
> versions of libtiff), it would be nice.
>
>  Regards,
> VZ
>
>
>   
>