[JDEV] Message timestamps

[Daniel Arbuckle]

I guess we need to clarify some terminology here. When I say a message was
sent at a given time, I mean that that time is when it was transmitted
from the senders client to the sender's home server. When I say a message
was received at a given time, I mean that that time is when the receiver's
home server got the message.

By using those numbers, the person receiving the message can know how long
ago the message was transmitted from the sender, or how long it has been
sitting in the delivery queue waiting for him to get back online.

The sent time is the more intuitively obvious one to use, except for the
spoofing problem.

The reason I care about people spoofing the sent time of a message is that
I am planning on having my client be able to prioritize incoming messages,
and one of the criteria for prioritization will be the timestamp, if there
is one. I don't want people to be able to  give their spam a high priority
by lying to the server.

There's also the problem of servers running on machines with mis-set
clocks.

On the whole, it seems better to have the receiver's home server do the
timestamping. There only needs to be a stamp for the time the message
arrives at the server, because the receiving client knows what time it is
when it gets the message. 

Daniel

On Mon, 4 Oct 1999, Steven Wagner wrote:

> I think there should be the following -
> 
> SENT - stamped by the receivers server when the msg is received
> RECEIVED - also stamped by the receivers server, but just before it delivered to the receiver.
> 
> Dan -  I think the receivers server should do the SENT time stamp also, rather then the receivers server.  This avoids
> the spoofing problem.  I don't see any other problems with this.
> Having the time stamp done by the server avoids the problem of users having the wrong date configured on their
> computer.
> 
> If we want to simplify things, the sent date is more important then the received.
> There is no reason why we would have to simplify this though.  Keep the protocol robust ... and we can just make the
> simple clients only display the SENT date as to not confuse the average user.
> 
> Daniel Arbuckle wrote:
> 
> > Regarding the location where timestamping is performed:
> > Obviously the timestamp should be inserted by the server; otherwise evil
> > clients could spoof the information. But it's not enough to say "the
> > server" should do it, because there are probably more than one server. If
> > the timestamp represents the time the message is sent, then the sender's
> > home server should apply the timestamp -- unfortunately, this also leaves
> > open the possibility of spoofing. If the timestamp represents the time
> > when the message arrived in your mailbox, then your own home server should
> > apply it to the message.
> >
> > The question is: "what does the timestamp represent?"
> >
> > Daniel
> >
> > On Mon, 4 Oct 1999, Steven Wagner wrote:
> > > Good point Scott.  Maybe we should consider which standard a programmer would be less likely to confuse or
> > > misinterpret.  There are obviously going to be many 3rd party clients in the future, probably some sloppy ones
> > > also if Corel is planning on making one, and we don't want the date getting messed up.
> > > Is the time date stamping going to be done on the server-side?  I hope so.  If it is, maybe my above point isn't
> > > as important.  :)
> > >
> > > Steven
> > >
> > > Scott Robinson wrote:
> > >
> > > > I would like to touch on processing time. It was mentioned in a previous
> > > > post that the ISO standard would be easier for a program to decode than the
> > > > RFC. I would note that once the C/C++ is written, the encode/decode times
> > > > are minimal if not non-existant.
> > > >
> > > > Scott.
> > > >
> > > > * Steven Wagner translated into ASCII [Mon, Oct 04, 1999 at 04:37:28PM -0700][<37F93A38.3E2DA99A at impulse.net>]
> > > > > As long as we avoid the problem that email has of email clients all having many
> > > > > different date formats.  Which one is it that does that?  Im confused.
> > > > >
> > > > > I think we should go with the date that is the most effecient for compatibility
> > > > > and readability by the code.  I don't think we should go with straight binary,
> > > > > but something that is more code effecient should be better then something that
> > > > > is more human readable.  The client is going to have to parse the date apart
> > > > > anyways so that it can display it in whichever way the user chooses (or are we
> > > > > not going to give them a choice?)
> > > > >
> > > > > I hope my input is valid.
> > > > >
> > > > > Steven
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > _______________________________________________
> > > > > jdev mailing list
> > > > > jdev at jabber.org
> > > > > http://mailman.jabber.org/listinfo/jdev
> > > > >
> > > >
> > > > _______________________________________________
> > > > jdev mailing list
> > > > jdev at jabber.org
> > > > http://mailman.jabber.org/listinfo/jdev
> > >
> > >
> > > _______________________________________________
> > > jdev mailing list
> > > jdev at jabber.org
> > > http://mailman.jabber.org/listinfo/jdev
> > >
> >
> > _______________________________________________
> > jdev mailing list
> > jdev at jabber.org
> > http://mailman.jabber.org/listinfo/jdev
> 
> 
> _______________________________________________
> jdev mailing list
> jdev at jabber.org
> http://mailman.jabber.org/listinfo/jdev
>