From: Charles Lindsey (chl@clerew.man.ac.uk)
Date: Fri Feb 13 2004 - 04:47:59 CST
In <87hdxw2c9e.fsf@windlord.stanford.edu> Russ Allbery <rra@stanford.edu> writes:
>Charles Lindsey <chl@clerew.man.ac.uk> writes:
>> Russ Allbery <rra@stanford.edu> writes:
>>> Then we don't and we make people who are doing complex gatewaying use
>>> some other mechanism of loop detection or bypass the normal posting
>>> software by using a special interface that allows preservation of
>>> Injection-Date (like, say, IHAVE).
>> No, we say that they MUST retain whatever Injection-Date header was put
>> there the first time it was injected. That was your original suggestion,
>> and indeed it stops all loops.
>You say "no" but you're agreeing with exactly what I said above. Complex
>gateways should use some other interface that allows them to preserve the
>Injection-Date header.
>Preserving existing Injection-Date headers for regular posting is highly
>unwise for all the reasons that we're talking about moving away from using
>the Date header.
Eh? We are going around in circles.
1. I initially proposed that the Injection-Date should normally be
rewritten by the injector if there was one already present on arrival.
2. I was roundly told by yourself and John Moreno that this was wrong,
would lead to loops, and it was essential to keep the old Injection-Date
in all circumstances.
3. I pointed out the case of the man with a news server in his laptop who
wandered around for three days trying to find a site to upload to. The
response was that there was no way for an injector to detect this as a
special case, and if his article failed to propagate as a result, then
that was Tough.
4. So I changed my position and agreed that an already present
Injection-Date should NEVER be overwritten (unless it was clearly in the
future or somesuch). That position is always safe (it never causes loops)
but may occasionally cause articles to propagate badly.
5. Now you are asking me to go back to position #1, and injectors are
somehow supposed to detect and deal with articles that come from "unusual
gateways", or we are to rely on the "unusual gateways" to know what they
are dpoing and to do it right, which seems risky to me.
> In off-line news reading situations where the person is
>running their own local news server, the Injection-Date may be
>significantly stale. Likewise for the workaround situations such as the
>one that John Stanley described. In those situations, I think the onus is
>on the agent doing unusual reinjection to strip the erroneous
>Injection-Date before reinjecting the message (an operation that should
>require manual configuration and therefore for someone to know what
>they're doing).
I think it would be much safer to put the onus on the man with the news
server on his laptop to remove the stale Injection-Date manually - since
it is only his article that will suffer if he fails to do so.
-- 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@clerew.man.ac.uk 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