Received: by 10.224.138.146 with SMTP id a18mr1341511qau.6.1343313751533;
        Thu, 26 Jul 2012 07:42:31 -0700 (PDT)
Received: by 10.66.77.41 with SMTP id p9mr1450281paw.8.1343313751312;
        Thu, 26 Jul 2012 07:42:31 -0700 (PDT)
Path: a15ni113198934qag.0!nntp.google.com!q21no1329092qas.0!news-out.google.com!p10ni64808639pbh.1!nntp.google.com!border1.nntp.dca.giganews.com!border4.nntp.dca.giganews.com!border2.nntp.dca.giganews.com!border2.nntp.ams.giganews.com!nntp.giganews.com!feed.xsnews.nl!border-1.ams.xsnews.nl!newsfeed.straub-nv.de!news-2.dfn.de!news.dfn.de!goblin3!goblin.stu.neva.ru!news.franciliens.net!news.muarf.org!nntpfeed.proxad.net!proxad.net!feeder1-2.proxad.net!74.125.178.16.MISMATCH!postnews.google.com!glegroupsg2000goo.googlegroups.com!not-for-mail
From: sinbad <sinbad.sin...@gmail.com>
Newsgroups: comp.unix.programmer
Subject: Re: order preserving cksum
Date: Thu, 19 Jul 2012 05:10:34 -0700 (PDT)
Organization: http://groups.google.com
Lines: 37
Message-ID: <9de0bc4d-7846-44a3-b52c-55d9fe2a4489@googlegroups.com>
References: <bc570559-7387-4221-8e4b-36db3d7693d6@googlegroups.com> <5007c48a$0$6130$426a74cc@news.free.fr>
NNTP-Posting-Host: 125.16.128.122
Mime-Version: 1.0
X-Trace: posting.google.com 1342699945 3239 127.0.0.1 (19 Jul 2012 12:12:25 GMT)
X-Complaints-To: groups-abuse@google.com
NNTP-Posting-Date: Thu, 19 Jul 2012 12:12:25 +0000 (UTC)
In-Reply-To: <5007c48a$0$6130$426a74cc@news.free.fr>
Complaints-To: groups-abuse@google.com
Injection-Info: glegroupsg2000goo.googlegroups.com; posting-host=125.16.128.122;
 posting-account=YM05VwoAAABfhqQvMhVh5BYqiwLHuZYv
User-Agent: G2/1.0
Bytes: 2787
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

On Thursday, July 19, 2012 1:55:47 PM UTC+5:30, Nicolas George wrote:
> sinbad , dans le message
> &lt;bc570559-7387-4221-8e4b-36db3d7693d6@googlegroups.com&gt;, a =E9crit=
=A0:
> &gt; i want to know if there is any cksum/hash function
> &gt; that can be used to apply cksum/hash on a set of data.
> &gt; for example if have the set { 23,45,67,89 }. if i compute
> &gt; the cksum/hash, no matter what order i compute it i should get
> &gt; the same result. does the current cksum functions already
> &gt; do that ?
>=20
> Be careful. Your subject line is about &quot;order-preserving checksums&q=
uot;. That
> means: for all input A, B, A &lt;=3D B implies checksum(A) &lt;=3D checks=
um(B). This
> is possible, but such a checksum is very weak, because a small change (a
> single corrupted byte) in the middle of the input will very probably not =
be
> detected.
>=20
> Your explanations rather speak of permutation-invariant checksums. That i=
s
> possible too, but to design it properly, you should probably be more,
> especially about the scale the permutations applies to: byte-level? recor=
d
> level, and if so, what is the size of a record?
>=20
> A simple way of achieving such a checksum is to sort the input before
> applying any checksum function. But this is costly, there are much more
> efficient solutions.

yes permutation-invariant checksums are what i am looking for.
any pointers to the efficient solutions ?

thnks
sinbad