[Imap-protocol] Mailbox root name
mrc at CAC.Washington.EDU
Mon Jun 25 08:15:15 PDT 2007
On Mon, 25 Jun 2007, Timo Sirainen wrote:
> > 1 LIST hello/world/ ""
> > It returns "hello/" as the mailbox root, even if hello doesn't exist.
> > Why? Does Pine really depend on this?
That's because that form of LIST is a syntax-only operation. What it's
telling you here is that the top-level of "hello/world/" is "hello/".
Pine is probably more likely to do
tag LIST hello/world ""
> > I remember I changed Dovecot to do
> > this a few years ago because Pine wouldn't work otherwise correctly with
> > ~/mail/ or mail/ or similar folder collection paths (I used to return ""
> > as root). Now that I'm looking over the code I'm just wondering why.
Uh-oh, you had better be careful here!!!!!!
~/mail is a very different case from mail/ because ~ serves as a hierarchy
breakout character. The root in
tag LIST ~fred/mail/ ""
is most certainly *not* ""! It is "~fred/".
In fact, that's probably what is going on here. Since ~fred/ is a
namespace reference, that form of LIST is actually used, whereas
tag LIST fred/mail/ ""
probably isn't actually used but is "fred/" for consistency. I don't dare
change it (see below).
> Looks like I can't reproduce this anymore. Maybe there were other things
> that caused Pine to break. So is it OK then if I return mailbox root
> namespace prefix, and if there is no prefix then just return ""?
I would recommend that you not do so. It is safer to you copy what UW
imapd does, even if it doesn't seem to make sense. I don't pretend to
understand or remember hierarchy in its full, hideous, glory.
You weren't around during the hierarchy wars (nearly 15 years ago); but
everybody who was has evil memories. I have the most evil memories of
all, as I was caught in the middle. Nobody was really happy with the
result; but what came out was a carefully-balanced compromise between two
very different views of how hierarchy and hierarchy navigation should
The underlying purpose of the reference argument in LIST is to provide the
information for a client to have the concept of "connected directory" even
though the server has no such thing. Basically, the reference is either
"I am connected to this directory" or "I am connected to the directory
that contains this name", depending upon if there is a trailing hierarchy
I forget why getting the root of the hierarchy was important, but it was.
If I were to alter the status quo, I am sure that someone will call me to
task for betraying the Great Compromise.... ;-(
-- Mark --
Democracy is two wolves and a sheep deciding what to eat for lunch.
Liberty is a well-armed sheep contesting the vote.
More information about the Imap-protocol