RE: MD5 and RPG or...?

Tom,

The Java program and shell script will match the output of md5sum run
on another system -- I wrote it to prove to our audit group that our
application repository stored off of the iSeries matches what is being
run in production. The check is now built into our build process so the
system (Gentoo Linux) that generates our applications and pushes them to
the iSeries stores an MD5 hash that can be read via an HTTP call. Before
unpacking on the iSeries we generate the same check sum and compare it
to the build machine. I expect that the cypher MI instruction may be
faster but the difference is unlikely to matter. A typical build package
on our system is about 20 Meg on our system and I just ran the md5sum
script I posted. It took about 2 seconds (on an 825). When I ran it over
a 500M database file it took about 40 seconds.

David Morris

>>> qsrvbas@xxxxxxxxxxxx 03/31/05 6:01 PM >>>
midrange-l-request@xxxxxxxxxxxx wrote:

>   7. RE: MD5 and RPG or...? (Wilt, Charles)
>
>You should be able to find more info in the archives.  But there's a
couple of parameters that allow you to call CIPHER repeatedly, once for
each chunk of an unlimited size file.

Charles:

There are a lot of archive references, but nearly all that I read were
concerned with individual calls to CIPHER. The only ones that seemed
beyond that were Leif's and the lone unanswered one (which is what made
me wonder how fruitful further searching/reading would be).

It'll be interesting to see how a series of CIPHERs matches with
md5sum.

Thanks.

Tom Liotta

-- 
Tom Liotta