Re: msgid
  By: Pascal Schmidt to Maurice Kinal on Fri May 07 2004 05:39 pm

 >  MK>           ^AMSGID: origaddr serialno
 > [...]
 >  MK>      The serial number may be any eight
 >  MK>      character hexadecimal number,  as long as it is unique - no two
 >  MK>      messages from a given system may have the same serial number
 >  MK>      within a three years.  The manner in which this serial number is
 >  MK>      generated is left to the implementor.
 > Note that it says "may be", there is no absolute
requirement for this. In fa
 > it goes on to say that an eight character hex number is *only* okay if it is
 > unique over a three year period. Plus nothing in the working forbids spaces 
 > the "serialno" field.
 > 
 > It also doesn't specify that the "origaddr" should be a
Fido address, becaus
 > then they wouldn't need to allow quoting in case there were spaces in there.
 > 
 > This standard isn't strict enough to allow any software making wild assumpti
 > except from the one that there'll be one space somewhere in the middle of th
 > MSGID. :)
 > 
 > Apart from that, the claim from that SBBS (was it that?) author that repeati
 > the same hex number in different casing makes a MSGID somehow *more* unique 
 > clearly bogus - if both numbers are always the same.

I realize this is an old thread, sorry for the late reply.

The reason for the repetition of the hexadecimal serialnumber (in uppercase the
second time) is to produce a longer source string for a "digest" alogorithm
(e.g. CRC-32, MD5, etc). Some Fido MTAs store "digests" of
MSG-IDs, rather than
the MSG-ID itself (and use the digests for later dupe-message detction).
Feeding the digest algorithm more text data is more likely to produce unique
digests.

In other words, the likely-hood of 2 different MSG-IDs producing the same CRC
is reduced when using longer MSG-IDs with more unique-to-that-message data.

                                                  digital man

Snapple "Real Fact" #50:
Mosquitos have 47 teeth.
--- SBBSecho 2.10-Win32