RE: ffpim revised draft

> At 08:11 AM 3/19/02 -0800, Dan Wing wrote:
> > > The TIMELY spec actually requires support for DSN (and more).
> >
> >Thanks.
> >
> > > (But it may help to be more explicit.)
> >
> >Yes, it would -- perhaps by pointing out that FFPIM achieves is
> >higher reliability of delivery confirmation by TIMELY's requirement
> >to implement DSN.  If there are other aspects of TIMELY that FFPIM
> >finds beneficial, especially if they're SHOULDs in TIMELY, it's vital
> >that FFPIM calls them out explicitly.  Currently, FFPIM only mentions
> >the timely delivery of TIMELY as the important part of TIMELY.
> 
> This was an illustration of how I read what I expected to find.  I 
> read the 
> text that Dave drafted, and thought it picked out quite nicely that the 
> thrust of TIMELY wasn't to make email faster, but to make it more 
> deterministic (my words).  But that's what I expected to read ;-)
> 
> [[[
>       Internet mail is often reliable and speedy.  However it
>       displays a very wide range of variability, depending upon
>       details such as software implementation, systems operation,
>       network connectivity and network activity.  By contrast, ...
>   ]]]

Wordy, but okay.

> Maybe, also add here:
> 
> [[[
>       But when the network is congested or software implementations are
>       poor, messages can be lost without any warning to the sender.
> ]]]

Congested networks don't, to my knowledge, cause failure of SMTP.  The
sender will continue trying to send the message for days and days
until the sender gets a positive acknowledgement.  During those 
multiple delivery attempts you could well have multiple deliveries,
of course (as documented a decade ago in RFCmumblemumble).

As for buggy software implementations causing problems -- buggy software
is everywhere, not just in Internet software!

> Other part of the text:
> 
> [[[
>       This specification uses a set of capabilities that permits
>       an originator to request that the email transport system
>       seek a particular timeliness in delivery and then assures
>       that the system will report the success or failure of that
>       request.
> 
>       The specification for this technique is in [TIMELY].
>       Implementations that are conformant to FFPIM MUST support
>       timely completion.
> ]]]
> 
> Seems to capture the intent when you know what to look for.  Maybe this?:
> 
> [[[
>       The specification for this technique is in [TIMELY], which
>       requires and extends support for [DSN] and [DELIVERBY].
> ]]]

That's perfect.  Thanks.

> (+ appropriate references)
> 
> It's been a while now, but I don't think there are any optional 
> (or SHOULD) 
> elements in [TIEMLY] that need to be mandated by FFPIM.

Okay.

Thanks, Graham.

-d