yeager@SUMEX-AIM.STANFORD.EDU (Bill Yeager's Mac) (02/16/90)
I can give a brief description of how we do this in the knowledge systems lab at Stanford. We use a server/client model for mail. Here the mail resides on the server, and is read/replied-to on the client but ALL updates to the mail file take place on the server. To accomplish the communication between the client/server we use IMAP(rfc1064) which is the Interactive Mail Access Protocol. This was principally designed by Mark Crispin when he was a member of our staff with lots of input from me and Rich Acuff. I wrote the "imapserver" which runs on unix based systems and is mail format independent with respect to imap which is similarly designed. The server supports mtxt(old tops20 mail.txt format now supported by Columbia mm on unix boxes) and what we will call "mbox" format for unix based mail. In this latter format the user's mail can stay in the spooled mail file or be moved to a file called mbox on the user's home directory. These options can all be set system wide or per user ... So, we also have an imap client, macMS, that is used in our lab by about 70 or so people is my guess. We are all pretty happy with it. This is also an client for TI Explorers(we do lots of lisp work around here) but have macs on everyone's desks as personal workstations. The mac mail client is now in beta-testing but not a lot of work is being done on it. Bill PS>POP and the like are proper subsets of IMAP. IMAP is transactionally based rather than message or mail file based which provides for lots of efficiency and good user throughput on reasonably loaded servers. -------