Re: Time stamps

On Wed, 17 Mar 1999, Daniel Stenberg <daniel.stenberg_at_sth.frontec.se> wrote:
>
> On Tue, 16 Mar 1999, Ashley M. Kirchner wrote:
> >
> > One thing I noticed is that hypermail has a tendency to grab the
> > top line to report the Last message date, while grabbing the Date line
> > in the headers to timestamp each message.
>
> Yeps.
>
> > This is what originally triggered my curiosity, why the
> > discrepancy. By looking at the headers, I now see which date is being
> > used for what. My question now is, why?
>
> Well I can't answer. I believe it was solved like this already in the 1.02
> and no one ever changed that. A guess would be that since the from-line is
> added by a server, and server's in general tend to have the clocks better
> set than clients, it would be a better reference.
>
> I'd prefer to use the Date:-string, it is more likely to reflect the time
> the send sent the mail which I think most often is the interesting date.
> That would of course also require a better date parsing function.

I agree with Daniel. I suspect this is the reason Kevin had the system use the date generated by sendmail (or other MTA) instead of the RFC821 Date: field. There is the annoying problem with the fact that sometimes people don't realize the clock on their PC needs reset, and the message gets sent from an MUA like Eudora with a date of Jan 1, 1980. Additionally, mail can be delayed in transit (e.g. a POP client delivers it to an SMTP server where it sits for a couple of days because a router is down), so the MTA generated date on the local system let's you know in what order the mail arrived at the server.

On the other hand, using the RFC821 Date: field is more likely to provide a date which adds context to a message. As we all know, it is possible to have a message dated the next day arrive as a reply to another message if the sender is across the international date line.

All this is part of the reason I suggested that it would be nice to have hypermail "normalize" the dates in some fashion (e.g. Zulu or GMT). It would make the sorting and arrangement a whole lot more reliable.

In order to solve the Jan. 1, 1980 problem, you could have hypermail do a simple check to see if the message date is *earlier* in time than the first message in a particular archive, or archive segment. If it is, you could have hypermail take some action like tagging the date as "unknown".

-- 

   Craig A. Summerhill, Systems Coordinator and Program Officer
   Coalition for Networked Information
   21 Dupont Circle, N.W., Washington, D.C.   20036
   Internet: craig_at_cni.org   AT&Tnet (202) 296-5098