Right. It only seems to break on this particular set of files
so far. I'll get a tar (18MB) on a web page this morning.

					Gary D. Duzan


p.s. I tried rebuilding libbz2 and bzip2 and got the same thing.


In Message <199911021229.XAA06180@balrog.supp.cpr.itg.telecom.com.au> ,
   Simon Burge <simonb@NetBSD.ORG> wrote:

=>Gary Duzan wrote:
=>
=>> Can anyone else recreate this sort of thing? It would be bad if
=>> bzip2 could systematically generate corrupt output. It could also
=>> be an Athlon CPU bug, I suppose, since that's what I'm running it
=>> on, but I tend to doubt that.
=>
=>I just tried a random selection of directories on NetBSD/alpha
=>and didn't get a problem.  Can you put the tar file of what breaks
=>somewhere I can ftp it so I can try the same data?
=>
=>Simon.