[OpenAFS] MD5 sums for Windows *msi files

Marcus Watts mdw@spam.ifs.umich.edu
Tue, 29 May 2007 16:33:09 -0400


Kim Kimball <dhk@ccre.com> writes:
> Hello,
> 
> I've tried three different md5 tools (Windows md5summer, md5sum 
> Solaris/RHEL) and cannot get the MD5 sums to match.
> 
> The sums I get are consistent across the tools I'm using.
> 
> Is there a mode switch I'm missing, or something like that?
> 
> Thanks!
> 
> Kim

What exactly are you trying to checksum?  What
URL did it come from?  Is the byte length exactly equal?

Off-hand, for source (text), crlf vs lf might break comparison.
Compressed vs. uncompressed may also be an issue.
Web browsers commonly do uncompression without asking.
Recompressing an uncompressed file may not produce
the same results as the original, even if the same
compression algorithm is used.

You might try checksuming "null" data to check your md5
implementations:
	$ md5sum /dev/null
	d41d8cd98f00b204e9800998ecf8427e  /dev/null
	$ 
For Windows, if "NUL" or \Device\Null doesn't work use an empty (0 length)
file.  If you do get different numbers, start by checking for byte swapping.
For instance, do you see 7e42f8ec98 ... ?  Supposedly this shouldn't
happen.

Besides md5sum, also try "openssl md5" on unix (and if possible windows).
Numbers should match, output may vary slightly in format.

Besides md5summer, there's also an unsupported microsoft tool
called "Microsoft File Checksum Integrity Verifier" that can apparently
compute md5 checksums (at least that's what wikipedia claims.)
Again, output format varies.

I doubt this is relevant, but if you really do have .msi files that
seem to be different, here's a .msi<->xml converter that might be interesting
http://msi2xml.sourceforge.net/

				-Marcus Watts