ncjuul@diku.dk (Niels Christian Juul) (12/29/90)
Apologies to everybody! Yesterday (December 27th, 1990) I submitted an article: 1>From: ncjuul@diku.dk (Niels Christian Juul) 2>Newsgroups: comp.object,comp.lang.smalltalk,comp.lang.c++,comp.lang.eiffel,comp.lang.objective-c,comp.lang.clos,comp.lang.modula2 3>Subject: Garbage Collection (Position Papers from ECOOP/OOPSLA'90 available) 4>Keywords: garbage collection, object-oriented, storage management, ftp 5> 6> xxxxx.xxx xxx xxxx. xxxx xxx 7>Message-ID: <1990Dec27.131916.5017@odin.diku.dk> 8>Date: 27 Dec 90 13:19:16 GMT 9>Sender: news@odin.diku.dk (Netnews System) a>Distribution: comp b>Organization: Institute of Computer Science, U of Copenhagen c>Lines: 111 The article was also cross-posted in several groups. The cross-posting may be dubious, but that's not the problem here. The article has been duplicated many times at different sites. ============================================================= We are very sorry about any inconvenience this may have caused. -------------------------------------------------------------- Our staff has inspected our news system closely, but has found nothing wrong on our sites. Our C-News seems to be fully compliant with the appropriate RFC's. The duplication seems to be caused by a legal but very stupid keyword field in the original header (line 4-6 above) which may have fooled some news systems So far we have seen no duplicates here, but reports say that up to 25 duplicates may be found on some sites. The problems seems not to grow further (I wish this will be true). It is of course a mess, for which we want to apologize. Niels Chr. Juul PS. So please remove me from your kill-file next year (:-). ------------------------------------------------------------------------------- Niels Christian Juul Email: ncjuul@diku.dk DIKU (aka Dept.Comp.Sci. Univ.Copenhagen) Phone: +45 31 39 64 66 ext.405 Universitetsparken 1 -- DK 2100 Copenhagen Direct: +45 31 39 33 11 -- 405 DENMARK Fax: +45 31 39 02 21 ------------------------------------------------------------------------------- Technical explanation: If you look at the header fields (on duplicates) you will find different message-id's, letting the duplicated messages originate from different sites. Our staff has tracked down the problem to: 1. The original submission had a stupid (but legal to RFC 822) keyword field, spanning multiple lines (and the second only contained a Horizontal TAB). 2. Until now we have observed the problem to originate from only 12 sites. Of these 4 runs ANU News (v. 6.0) the other sites are unreadable from here. 3. Some systems out there seem to have interpreted this line as the end of the header, thus they did not find the original Message-Id field placed lower in the original header. Following another RFC, articles must have a message-id in their headers. So these systems may have 'thought' that this was a new article and re-posted it with their own message-id. This may be done as many times as these sites recieve the original message. The net result is that as long as the original article is received by these 'non-RFC 822 compliant' systems a new duplicate may be generated. Thus we are not able to cancel these duplicates nor to prevent new duplicates from being generated. However the path field of the original article seems to be reused, thus sites which filter out articles that has been on their site previously (determined by the site being in the path), should only receive dublicates generated up- streams in the news feed.