[imp] Deleted messages not moved to trash from client
Joseph Brennan
brennan at columbia.edu
Fri Apr 10 14:30:30 UTC 2009
Ernest John Messersmith III <ejm at ejm3.net> wrote:
> In the
> case of my iPhone and Webmail (when properly configured), when I delete a
> message using its interface it is actually sending a COPY (to Trash
> mailbox) command, followed by an EXPUNGE (from source mailbox) command.
> Whereas the other clients (Outlook, Outlook Express, Windows Mail) are
> simply sending a MARK AS DELETED command.
First of all the trash hack and simple delete are both available in
nearly every client. However since there is no standard for the trash
hack, clients call it whatever they choose to call it. I see most
commonly trash, Trash, and Deleted Messages. Maybe non-English clients
call it words in other languages. Notice that if you use more than one
client, you might have more than one trash folder, unless you configure
them to use the same name, if you can do that. IMP allows it.
Second, the client sends COPY message, DELETE message, and EXPUNGE source
mailbox, and then sometime later when you empty trash, it has to SELECT
trash, DELETE message, and EXPUNGE trash. Gaaah!
> Now in the case of Gmail since clients connect to a different port I'm
> going to assume there is an intervening layer they are providing which
> straightens all this mess up and behaves in one consistent way no matter
> what junk a client sends. This is where protocol messages get
> "translated" if you like.
I have no idea how that could be done. How would it guess that you use
the trash hack at all and what name you call the trash folder? Suppose
I don't use the trash hack, but I decided to move messages about trash
to a folder I named trash? That would look the same on the server side.
(The example is inspired by the time a professor stored mail about the
core curriculum in a file named core, which some helpful cleanup script
kept wanting to delete for him!)
Joseph Brennan
Lead Email Systems Engineer
Columbia University Information Technology
More information about the imp
mailing list