On Wed, Feb 11, 2004 at 01:22:56PM -0800, Dave Crocker wrote:
HD> I'd propose to invent "Mail Cookies" similar to
HD> HTTP cookies. Could allow to follow thread and
HD> to identify replies to former messages.
1. How is that different from the Message-ID of the first message in
The Message-ID is just a relation to the first message in a thread.
It is not a relation to a particular domain or mail address.
The thread-Message-ID is useful only as long as you reply to
another message of the same thread. When you leave the thread or
do send a new message, it's gone. Furthermore, it is sent to
everybody on the same thread. This mechanism is intended to
allow the MUA to sort the messages into threads and display them.
A cookie is a different thing, even if it might look similar on
a first glance. Once it is stored in the MUA (similar to cookies
stored in a Webbrowser), it would be included in the header of
any new message sent to a particular address. In contrast to
MessageIDs it can carry information. The MessageID is like
a random number: It's intended to be unique, but not a container for
When a message comes in with a cookie, this might e.g. affect
the way the message is treated, maybe given higher priority,
routed differently. Maybe a simple authentication method for
mailing lists. Maybe a hint for sorting messages into folders.
Maybe to automatically include your customer number for mails
sent to any company.
2. Threads often develop sub-structure, looking like a tree. How would
the subordinate branches be distinguished?
Not at all. Cookies are not a different word for the MessageID, and
they are not intended to be a sorting criteria for branches etc.
They have nothing to do with threads and branches.
I send my customer a cookie, i.e. in a message I have a header
set-cookie: *(_at_)danisch(_dot_)de valid until 2008 customer 12345
Then (if the MUA supports and the customer agrees) any message from
the customer to my domain, even new messages, would contain a cookie: