solomon@uwvax.UUCP (06/19/83)
One of the sites philabs, linus, genrad, decvax, or harpo is fiddling with message ID's, causing duplicate items. I got two versions of an item in net.auto. They took the same route here except that one of them went directly from harpo to seismo while the other took a detour through the above-named sites. The second copy had several header lines deleted (References, Organization, and Relay-Version) and had a ".UUCP" ADDED to the Message-ID. The change to Message-ID meant that it was not detected as duplicate. I suspect that some site is running an old version of news that strips these fields, and another site is regenerating Message-ID from Article-I.D. using a different algorithm from the one used at the source of the item (uw-june). Any comments? Diffs between the two copies follow: 3,4c3 < Posting-Version: version B 2.10 5/3/83; site uw-june < Path: crystal!uwvax!seismo!harpo!floyd!vax135!cornell!uw-beaver!uw-june!emma --- > Path: crystal!uwvax!seismo!philabs!linus!genrad!decvax!harpo!floyd!vax135!cornell!uw-beaver!uw-june!emma 8c7 < Message-ID: <482@uw-june> --- > Message-ID: <482@uw-june.UUCP> 12,15c11,12 < Date-Received: Fri, 17-Jun-83 03:23:31 CDT < References: hou5e.560 <561@hou5e.UUCP> < Organization: U. Washington, Computer Sci < Lines: 9 --- > Date-Received: Sat, 18-Jun-83 01:28:37 CDT > Lines: 10 -- Marvin Solomon Computer Sciences Department University of Wisconsin, Madison WI solomon@uwisc ...!seismo!uwvax!solomon
chris@umcp-cs.UUCP (06/20/83)
Might it not be a good idea to have the history-check routines strip
off domain specifications (.UUCP, .ARPA) in IDs? Probably most of the
duplicate news items would then go away.
- Chris
--
UUCP: {seismo,allegra,brl-bmd}!umcp-cs!chris
CSNet: chris@umcp-cs
ARPA: chris.umcp-cs@UDel-Relaysmk@linus.UUCP (Steven M. Kramer) (06/20/83)
I really don't think it is linus. Genrad has the same software, so if this
was the problem with us, it would be with them also. My guess would
be harpo, since it seems to be the point where both copies are first found.
--
--steve kramer
{allegra,genrad,ihnp4,utzoo,philabs,uw-beaver}!linus!smk (UUCP)
linus!smk@mitre-bedford (ARPA)