RE: Generate hash code for a source member? -- MI400

Dan,

What I meant in my original post was simply use a data structure where
positions 1 thru 96 (or whatever) are the source records and 1 thru 96
are also the array.  So each byte is redefined.

If you are actually starting to think this might be useable, you could
use a trick I've used for generating hash values.  Overlapping the
array definitions like:

1 4 el,1
2 6 el,2
4 8 el,3

So each byte ends up in 2 elements.  That would catch any 2 character
transposition for sure.  And if you simply added the element number to
the element before cross footing, that should catch any "moving" around
of the data.  Then perhaps add the record number to the record total.
That would catch the moving around of records.

This might be workable.  As I said, it is brute force.   Not very
elegant.  But might be quick to program.

Bob

-----Original Message-----
From: mi400-admin@midrange.com [mailto:mi400-admin@midrange.com]On
Behalf Of Dan Bale
Sent: Thursday, May 09, 2002 5:02 PM
To: mi400@midrange.com
Subject: RE: [MI400] Generate hash code for a source member?

Ya know, I've used CMPPFM enough to realize that there was some special
stuff going on for it to be able to identify statements that are not
the
same, but still able to "match" them when the new file's statement was
changed from the old.

Also, I've found that most people don't realize that there is a
companion to
CMPPFM called MRGSRC (Merge Source Physical File).  I've only used it a
few
times, a long long time ago; could never get comfortable to trust it,
always
had to verify what it did and that took more time than if I just did
the
"merge" myself.  But still, pretty neat stuff.

Tom, I understand that 20 bytes of hash is never going to absolutely
guarantee that two generations of the same hash value mean that the two
source members are identical.  Still, 20 bytes = 16**20 =
1,208,925,819,614,629,174,706,176 --- I think those are pretty good
odds.

Also, I had to re-read Bob Crother's original suggestion after Tom
clarified
something for me.  I  misinterpreted the "Read the source record in as
an
array of 32 bit integers."  Now I'm not sure what that really means.
How do
I turn a source record into an array of 32 bit integers?  Would this be
"better"/more unique than the CIPHER hash that Leif's MIFHASH app will
generate?

- Dan Bale
(I am *NOT* "Dale"
http://archive.midrange.com/midrange-l/200105/msg00281.html )
SAMSA, Inc.
989-790-0507
DBale@SAMSA.com <mailto:DBale@SAMSA.com>
  Quiquid latine dictum sit altum viditur.
  (Whatever is said in Latin seems profound.)

_______________________________________________
This is the MI Programming on the AS400 / iSeries (MI400) mailing list
To post a message email: MI400@midrange.com
To subscribe, unsubscribe, or change list options,
visit: http://lists.midrange.com/cgi-bin/listinfo/mi400
or email: MI400-request@midrange.com
Before posting, please take a moment to review the archives
at http://archive.midrange.com/mi400.