[Gdal-dev] Re: VC7 build log

[Frank Warmerdam]

Chapman, Martin wrote:
> Frank,
> 
> You will only lose precision if what's in the size_t is bigger than an
> int.  Are you storing anything bigger than that in the size_t?  If not,
> turn off the warning.  If so, perhaps you can shift the bits manually
> from a size_t to the int where needed.  Then you could keep the warning
> turned on and convert it cleanly.  Again, this warning is not applicable
> in your case it sounds like and that's why it's a warning and not an
> error.

Folks,

OK, Roger's project will be without the 64 bit compatability checking.

I would note that I am generally aware of 64 bit issues.  GDAL works fine
on several different 64bit operating systems (Solaris, IRIX, OSF/1) and
there is a special 64 bit file access API.

I have every intention of working on Windows systems on 64 bit architectures
when they become common enough for me to get access (Itanium?  AMD's 64 bit
chip?).  However, things will be limited to accessing chunks of no more than
2GB at a time as long as I use int's for size_t but access to very large
files is still fine.  Offsets via the large file API uses a type that is
64 bit on appropriate architectures for instance.

Thanks for digging up the notes Martin.  It helps explain why I didn't run
into these warnings with my makefiles too.  One thing that bugs me about
project files is that so many users of them are blissfully unaware of the
compiler options they are using.

Best regards,
-- 
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up   | Frank Warmerdam, warmerdam at pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush    | Geospatial Programmer for Rent