> (Short term) solution on Internet tends to be widespread as legacy > which will hardly obsoleted without fundamental flaws which prevent > users from using it anymore.
Of course, but such a transition would go unnoticed if done correctly. Example: A special SMTP handshake wich would be ignored by old implementations (servers or clients) and acted upon by newer software. The end user wouldn't notice anything but an improvement in performance on some servers. When the special handshake would go into effect, the sockets on both sides would go from being line-based to becoming transparent. The protocol could be based on single byte commands, length dwords and such, to allow for content independent encapsulation with a non-existant encoding overhead. New servers would use Unicode for internal representation of the plain text content of messages. Whenever mail would be exchanged with an old server, one way or the other, the new server would do the nessecary encoding/decoding and/or encapsulation. The first servers to be upgraded would barely use the new protocol. The handshake would simply not generate a reply at any server or client. This has to be expected and accepted during the transition from one protocol to another. The effect of the new protocol would be the gradual improvement in effective bandwidth usage on the Internet. Another idea would be the addition of compression. One (or the implementation) could choose to compress individual message parts, individual messages or the entire content of the inbox of a mail account. This would also improve bandwidth usage, and have a noticable effect on servers exchanging large amounts of mail. I can only see advantages. -- Thor Harald Johansen
