markw@gvlf1-c.gvl.unisys.com (Mark H. Weber) (09/25/90)
After extensive investigation, I think I understand how the "Distribution: " header is supposed to work in conjuction with the "Newsgroups: " header, but I'd like someone to double check me. In order for a news article to be received by a system, or sent to the next system, there must be a matching entry in the "sys" file for for both the "Newsgroups: " and "Distribution: " headers. An empty or missing "Distribution: " header matches "world" in the sys file. So far so good? In Bnews, the Newgroups and Distributions are combined in the same field, so that a site which wants a full mainstream news feed would have a sys file entry something like this (I'm in Pennsylvania, USA): mysite:comp,sci,rec,talk,news,soc,misc,world,na,usa,pa:: In Cnews this field can be separated into separate Newsgroups and Distributions by use of a slash: mysite:comp,sci,rec,talk,news,soc,misc/world,na,usa,pa:: or mysite:comp,sci,rec,talk,news,soc,misc/all:: Am I still on the right track? Where this starts to get weird (and I become less self-assured) is when you get into the alternate and regional newsgroup hierarchies such as "vmsnet", "ba", "ca" or (in my case) "pa". Many articles which are posted to these groups have the hierarchy name in both the Newsgroup and Distribution header. Therefore: 1) I assume that in Bnews, the same field in the sys file is scanned twice for both headers, so that a single occurrence of the name ("pa" for example) matches both headers and allows these articles to be propagated. 2) I further assume that Cnews works the same way as #1 above if the field in the sys file has not been separated. 3) And finally, I assume that at Cnews sites where the fields have been separated, the separate fields are scanned uniquely, and in order for messages in the "pa" newsgroups which have been posted with "pa" in both headers to be propagated, there must be a "pa" entry in both fields, or a "pa" in the Newsgroups part, and an "all" in the Distributions part of the field (or vice-versa). Is this correct? -- Mark H. Weber | Internet: markw@GVL.Unisys.COM Unisys - Great Valley Labs | UUCP: ...!uunet!cbmvax!gvlv2!markw Paoli, PA USA (215) 648-7111 | ...!psuvax1!burdvax!gvlv2!markw
henry@zoo.toronto.edu (Henry Spencer) (09/25/90)
In article <960@gvlv2.GVL.Unisys.COM> markw@GVL.Unisys.Com (Mark H. Weber) writes: > 3) And finally, I assume that at Cnews sites where the fields have been > separated, the separate fields are scanned uniquely, and in order > for messages in the "pa" newsgroups which have been posted with "pa" > in both headers to be propagated, there must be a "pa" entry in both > fields, or a "pa" in the Newsgroups part, and an "all" in the > Distributions part of the field (or vice-versa). Correct. -- TCP/IP: handling tomorrow's loads today| Henry Spencer at U of Toronto Zoology OSI: handling yesterday's loads someday| henry@zoo.toronto.edu utzoo!henry
markw@gvlf1-c.gvl.unisys.com (Mark H. Weber) (09/26/90)
In article <960@gvlv2.GVL.Unisys.COM> markw@GVL.Unisys.Com (Mark H. Weber) writes: >After extensive investigation, I think I understand how the "Distribution: " >header is supposed to work in conjuction with the "Newsgroups: " header, >but I'd like someone to double check me. Thanks to tale@cs.rpi.edu (David C Lawrence), who sent me an excellent summary of not only the mechanical quirks of the two headers, but also some of the current thinking on the subject. I think I'm up to speed now. Thanks, Mark -- Mark H. Weber | Internet: markw@GVL.Unisys.COM Unisys - Great Valley Labs | UUCP: ...!uunet!cbmvax!gvlv2!markw Paoli, PA USA (215) 648-7111 | ...!psuvax1!burdvax!gvlv2!markw