JEHAD - Threading observation - Attn: Sherlog

[pls.se...@my.sig]

I've just seen my first case of "thread" collision, and it's in this
group.

As the number of messages in my database changes, these four
A.
> Subject: Auto-Expanding threads?
> From: rajs...@execpc.com
> Date: Thu, 03 Apr 1997 20:48:27 GMT
> Message-ID: <3345177b...@news.execpc.com>

B.
> Subject: Re: Auto-Expanding threads?
> From: pls.se...@my.sig
> Date: Thu, 03 Apr 1997 21:30:35 GMT
> Message-ID: <3351214e...@netnews.worldnet.att.net>
> References: <3345177b...@news.execpc.com>

C.
> Subject: Minor cosmetic problem?
> From: ta...@lava.net
> Date: Mon, 26 May 1997 16:51:19 -1000
> Message-ID: <338b4b86...@news.lava.net>

D.
> Subject: Re: Minor cosmetic problem?
> From: pls.se...@my.sig
> Date: Tue, 27 May 1997 03:17:15 GMT
> Message-ID: <33954ebb...@netnews.worldnet.att.net>
> References: <338b4b86...@news.lava.net>

sometimes thread as
>> A,B
>> C,D
sometimes as
>> A,B,D
>> C
and sometimes as
>> A
>> C,B,D

When I have time, I'll switch to subject sort and see what happens.
--
Jim [mailto:JLBradley#worldnet.att.net]
Checked [Free] Agent help and need more? http://sd.znet.com/~lance/
Want an answer fast? http://www.dejanews.com/forms/dnsetfilter.html
Internet beginner? http://www.which.net/nonsub/sitemap/using.html
More help with binaries? http://shell.ihug.co.nz/~ijh/index2.html

[pls.se...@my.sig]

;On Tue, 27 May 1997 10:29:18 GMT, fur...@sha.ubeK.qr (Sherlog) wrote:

>Table of collision probabilities for
>Forté's hash

>| 0.0001 0.001 0.01 0.1 0.5@ 0.9 0.99 0.999 0.9999
>|-----------------------------------------------------------------------------------
>| 67108786: 117 367 1162 3761 9646@ 17580 24861 30448 35157

The key observation here is that Forté's algorithm is likely to make a
hash (pun intended) of the threading in a group/folder well before it
fills up. My previously stated tolerance limit for this blunder is so
far to the left in the above table (367 messages) that I'm now looking
seriously at installing your CRC patch.