|Diff Tools email@example.com (2001-03-26)|
|Re: Diff Tools firstname.lastname@example.org (Richard Norman) (2001-03-27)|
|Re: Diff Tools email@example.com (Daniel Dunbar) (2001-03-27)|
|Re: Diff Tools firstname.lastname@example.org (Hans-Bernhard Broeker) (2001-03-27)|
|Re: Diff Tools email@example.com (Dennis Yelle) (2001-03-27)|
|From:||Hans-Bernhard Broeker <firstname.lastname@example.org>|
|Date:||27 Mar 2001 23:44:36 -0500|
|Organization:||Aachen University of Technology (RWTH)|
|Posted-Date:||27 Mar 2001 23:44:36 EST|
In comp.compilers email@example.com wrote:
> I've encountered a documented problem within Visual C++, that the
> same code will produce different sized executables, when compiled at
> different times or on different machines.
Hmm... MS sure has a strange idea of what 'documenting a problem'
means if they don't tell what this size difference really is coming
> Now, the problem is, my company wants to be able to see the
> differences between these two executables. If they are just
> time/date stamps, that's fine, but if there is other stuff (memory
> contents, etc) there may be problems, and we have to be able to
> determine that.
I'm not sure how well this really works with VC++-generated files, but
using Unix-like tools (also available on MS Windows, thanks to the GNU
and Cygwin project), I'd do the following:
objdump -x ver1.exe > ver1.dump
objdump -x ver2.exe > ver2.dump
diff -uw ver1.dump ver2.dump > diffs
and then have a look at the diffs. Executables with at least minimal
debugging information might be preferrable. Some post-filtering of the
.dump files might also be needed, to screen out some types of
differences. Comparisons of .obj/.o files might also be helpful to
check whether the difference happens at link or compile time.
Hans-Bernhard Broeker (firstname.lastname@example.org)
Return to the
Search the comp.compilers archives again.