What's the idea behind the png_constcast() macro implementation?

Hi,

among other platforms we are using libpng on IBM iSeries. After 
upgrading from an older version to libpng 1.6.34 we are getting crashes 
in libpng code. It looks like the following construct in pngpriv.h is 
responsible for the crashes:

#  ifdef _WIN64
#     ifdef __GNUC__
          typedef unsigned long long png_ptruint;
#     else
          typedef unsigned __int64 png_ptruint;
#     endif
#  else
       typedef unsigned long png_ptruint;
#  endif
#  define png_constcast(type, value) ((type)(png_ptruint)(const 
void*)(value))

On IBM iSeries it is not possible to cast pointer types to integer types 
and back to pointer types. See for example here:

https://www.ibm.com/support/knowledgecenter/SSAE4W_9.6.0/com.ibm.etools.iseries.pgmgd.doc/cpprog448.htm

The question is, what is the idea behind the intermediate cast to 
png_ptruint in the png_constcast() macro, before the (type) cast is 
applied? I did a quick test with gcc on Linux and Visual Studio on 
Windows, and leaving out the (png_ptruint) did not cause a warning.

Thanks
Stephan

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot