Re: Path header field

["Charles Lindsey" <chl@xxxxxxxxxxxxxxxx>]

In <59ftJMHK26mFFAKd@xxxxxxxxxxxxxx> Richard Clayton <richard@xxxxxxxxxxxxxx> writes:

>what I'd add to that now is that any site with multiple machines doing
>peering, handling injections etc is very likely to have its own custom
>schemes for ensuring that databases are synchronised (or indeed there
>may just be a central store of data). These are unlikely to depend on
>scanning Path header fields, but of course they may.  However, when
>articles depart from these systems to the rest of the world they'd like
>to add a site-wide identity to show that they have been dealt with, so
>as to discourage them being offered again. Hence the addition of
>multiple path identities by such machines...

>hence we see

>        !newsfeed.gamma.ru!Gamma.RU!
>        !supernews.com!corp.supernews.com
>        !nx01.iad01.newshosting.com!newshosting.com
>        !news.tele.dk!news.tele.dk!small.news.tele.dk

>and doubtless many more (those were found in seconds in a current
>feed...)

Essentially, we need to allow sites to insert more Path entries than the
number of hosts the article actually passed through, just as we need to
allow them to insert fewer than that number of entries if the internal
structure of the site does not need to be exposed outwardly. Just so long
as the leftmost one is the one most likely to be recognized by their
peers, so that MISMATCH checking proceeds smoothly. What is good for
MISMATCH checking is not necessarily best for preventing articles being
sent to them more than once.

>>and can we please then incporate that into the examples somewhere. 

>I'm not sure it's to be encouraged for systems that are not so complex
>as major peering hubs.  Mind you, I sometimes wonder if there's going to
>be all that much left soon that isn't a major peering hub.

I think there are all sort of sites out there with special circumstances
that nobody else is aware of. So if we included an example, we would
probably say something like ...!news.demon.net!demon!incoming.demon.net!...
with and explanation like "although the site news.demon.net does not
actually include a host the with name 'demon', it included that
<path-nodot> because it was aware of a large number of (client) sites that
still knew it by that name". But change "demon" into some other foobarish
name, of course.

Interestingly, we have been talking a lot recently about home sites that
included their own personal newsserver. Demon is unusual in that it
possesses an unusually large number of such clients, since it initially set
all its clients up that way.

-- 
Charles H. Lindsey ---------At Home, doing my own thing------------------------
Tel: +44 161 436 6131 Fax: +44 161 436 6133   Web: http://www.cs.man.ac.uk/~chl
Email: chl@xxxxxxxxxxxxxxxx      Snail: 5 Clerewood Ave, CHEADLE, SK8 3JU, U.K.
PGP: 2C15F1A9      Fingerprint: 73 6D C2 51 93 A0 01 E7 65 E8 64 7E 14 A4 AB A5