Re: #1416 Injection-Date: current wording proposal (corrected)

[Russ Allbery <rra@xxxxxxxxxxxx>]

Harald Alvestrand <harald@xxxxxxxxxxxxx> writes:

> There are 3 cases:

> - Message is singly injected. Injection-Date: does no harm.
> - Message is multiply injected, Injection-Date: on all copies are quite
> close to each other, because they all got injected in rapid
> succession. Injection-Date: does no harm.
> - Message is multiply injected, and some of the injections got
> significantly delayed.

Right, the problems only potentially happen in the third case, but that
third case includes many serial injection cases.

> I think most user agents will add Date: and Message-ID: by
> default. (Anyone have data on this, rather than supposition?)

Charles had some data, I believe.  If I remember it correctly, "most" is
iffy, but many do.

> So the net result of catering for the minority third case is that *most*
> messages on the Net will be sent without an Injection-Date.

Correct, until posting agents are updated to generate it.  This is the way
that I think it should work, since all the benefits of Injection-Date are
born by the posting agents, not by the injecting agents (which don't care
at all).

> If so, what's the point of claiming that we're "mandating" it?

We're mandating checking it instead of Date so that reading agents can
start using the feature.

-- 
Russ Allbery (rra@xxxxxxxxxxxx)             <http://www.eyrie.org/~eagle/>