IMAP (02 Feb 2003)

I've had cause to read the IMAP RFC rather a lot recently and it's a pretty good example of what not to do when deigning a protocol.

For example, in the FETCH command:


Do they really think that people are going to be using raw IMAP with a telnet client so often that these shortcuts are going to be useful? It's not exactly damaging, it's just stupid and a clear indication that the designers didn't understand protocol design

The point of protocol design it's to get all the required functionality in there. It's to do that with the minimum number of primitives

Now sometimes those primitives are pretty large and performance dictates that they shouldn't be broken down any further. But IMAP certainly cannot claim that it has hit that barrier.

Also, IMAP servers are supposed to parse MIME and present a breakdown to the client, the point being that the client doesn't have to fetch the whole message. Desirable functionality, but commands for fetching and substring searching in a byte range would allow clients to do that and not burden the server with the very-much-client-side-total-mess that is MIME. And this type of thing litters the IMAP protocol.