[Lula] md5sum weirdness on firefox

[Joe Bayes]

Hi folks,

This is a matter of curiosity, not system stability, but if you have
some spare time on your hands maybe you can help me figure out this
puzzle. 

The short version:

I'm running Fedora 6. I downloaded firefox-1.5.0.9-3.fc6.i386.rpm from
http://download.fedora.redhat.com/pub/fedora/linux/core/updates/6/i386/. 
When I tried to install it, yum complained about the md5sum of one
of the files. The Nice Folks At Redhat think it was a corrupted
download. 

I don't think the download was corrupted, and here's why: I downloaded
it several more times, and each time I got an identical, non-working
file. I downloaded the file from redhat directly to a third computer
(a Solaris box that isn't administered by me), then uploaded the
non-working file from my computer to the Solaris box, and compared the
files: they were identical.

However, on the Fedora box on which I'm trying to install firefox:
spoo:~/ff$ md5sum firefox-1.5.0.9-3.fc6.i386.rpm 
c1a703baffce31371ce91d2b830220d7  firefox-1.5.0.9-3.fc6.i386.rpm

On the Solaris box:
occs:/tmp/jbayes/thru_spoo$ md5 firefox-1.5.0.9-3.fc6.i386.rpm 
MD5 (firefox-1.5.0.9-3.fc6.i386.rpm) = 1284459a13f9898487f06ee925876b2c

On the computer belonging to the Nice Folks At Redhat:
[matej at hubmaier ~]\$ md5sum firefox-1.5.0.9-3.fc6.i386.rpm 
1284459a13f9898487f06ee925876b2c  firefox-1.5.0.9-3.fc6.i386.rpm

My question: Why do the checksums differ? If your answer is "because
the files differ", then how is it that the file transferred from
redhat to spoo to occs has the same checksum as the one on hubmaier at
redhat? 

Further details are available at
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=228915 .

Firefox-1.5.10 is already out, and it works just fine, so this is just
a matter of curiosity, not system stability. 

Joe

--
Joe Bayes -- jbayes at spoo.mminternet.com