Re: #1416 Injection-Date - Summary of options

["Charles Lindsey" <chl@xxxxxxxxxxxxxxxx>]

In <JF7Fto.G8J@xxxxxxxxxxxxxxxx> "Charles Lindsey" <chl@xxxxxxxxxxxxxxxx> writes:

>I think we have now said everything that is to be said on this, so I have
>produced a summary of where I think we have got to.

>It lists rules for what various agents are supposed to do with
>Injection-Date in various situations. Most of these are, I think, agreed
>(and they differ quite a bit from our starting point in the two versions
>of USEPRO and in assumptions made at the start of this thread, so that at
>least indicates some progress).

I had hoped for some response to this, especially from Russ, particularly
as to whether the points I claimed to be "agreed" were in fact agreed.
That would then enable us to proceed to the main issue of disagreement:

>There still remains the two conflicting possibilities (labelled IC and IR
>in this summary) that we have to choose between, though I think we
>understand the consequences of each better than we did.

I understand that Ruus has been tied up on other matters since then, but
that he hopes to return to active participation shortly.

In the meantime, however, I have encountered one feature that we need to
look at further. In addition to delays caused by a poster who prepares his
article offline and then delays for some days before injecting it, there
is also the case where the poster injects his article promptly but,
because it is posted to a moderated group, the moderator may then delay
for several days before approving it.


>Posting agents:
>---------------

.....

>P4. MAY include Injection-Date indicating time of actual injection (as
>opposed to composition).

...........

>P6. When injecting at multiple injecting agents, MUST include Injection-Date
>which MUST then indicate time of earliest injection at any such agent.


But of course all those multiple injections will now end up being sent to
the same moderator, and the Injection-Date provided by the posting agent
(which usually would not be aware whether a particular group was moderated
or not) is largely meaningless.

I am assuming, of course, that if the 'moderator' is actually a team of
moderators with some bot to distribute articles amongst them, they will
have set up arrangements to ensure that a given article only gets approved
once.


>Injecting Agents:
>-----------------

......

>I4. MUST NOT insert (or alter or delete) Injection-Date if it is already
>present.

So I think, in this case, the moderator MUST now discard any
Injection-Date already provided, and then proceed to inject the Approved
article in the normal manner, at which point he MAY then add an
Injection-Date, and MUST add one if he is injecting at multiple sites.

That will give the best chance of the article propagating normally, even
if the moderator delayed before Approving it.

There is a separate but related question of whether, for the benefit of
legacy injecting and relaying agents, he might need to do something about
a Date header that had been rendered unusually stale because of hiw own
delay. Usepro-06 made some special provision for this, and I think
Usepro-07 does also. Normally, of course, Date should be left strictly
alone as being a matter to be determined by the Poster, but one can hardly
blame the Poster for subsequent delays caused by the moderator.

Apart from this one case, I think all the other rules listed in my
original message are unaffected.

-- 
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