DEFINITIONS:
------------

The terminus "mailbox" is used throughout the code in accordance with
c-client speak.

What c-client calls "mailbox" is commonly referred to as a mail FOLDER.

In mailsync a box containing multiple folders is described by a "Store"
- have a look at store.h for details.

MSINFO format description:
--------------------------

ATTENTION: the msinfo format can change without warning!

The msinfo file contains synchronization information for channels as
defined in the configuration file. Each channel contains the mailboxes
associated with it and the message ids of all the mails that have been
seen in those mailboxes the last time a sync was done.

The msinfo file is a normal mailbox.

Each email therein represents a channel. The channel is identified by
the "Subject: " email header.

The body of an email contains a series of mailbox synchronization entries.
Each such entry starts with the name of the mailbox followed by all the
message ids that were seen and synchronized. Message ids have the format
"<xxx@yyy>" or "<>, tag: <value>".

Example:

From tpo@petertosh Fri Oct  4 12:18:13 2002 +0200
From: mailsync
Subject: all
Status: O
X-Status:
X-Keywords:
X-UID: 2095

linux/apt
<1018460459.4617.23.camel@mpav>
<1027985024.1352.19.camel@server1>
linux/mailsync
<001001c1d37a$39ee3d70$e349428e@ludwig>
<002b01c1e95c$94957390$e349428e@ludwig>

From tpo@petertosh Sat Nov 30 18:21:55 2002 +0100
Status:
X-Status:
X-Keywords:
From: mailsync
Subject: inbox

INBOX
<>, md5: <2292e5bda62ae83acbd11f3d61d62c7b>
<>, md5: <24c5f14bd1b8198520736843f369f415>

We have here two channels. The first is use for synchronization
of all my mailboxes and the second one solely for my inbox.

The first channel contains the mailboxes "linux/apt" and "linux/mailsync".
Each of those mailboxes has had two mails in it. The first is being
synchronized through the standard method, by using Message-IDs to uniquely
identify messages and the second one is using md5 checksums.

Take care - mailboxes with *empty* names *are* allowed.

RELEASE PROCESS
---------------
1. * release CVS tarballs
2. as soon as it seems stable:
   * make note in NEWS about alpha release
   * release alpha
   * make dist and upload
3. after some downloads without complaints:
   * verify that problems that the release was supposed to
     fix are actually fixed with the people that had them
   * update version number to beta in configure.ac
   * make note in NEWS about beta release
   * release beta
   * make dist and upload
4. after enough time:
   * update version number in configure.ac
   * make note in NEWS about release and beta period
   * tag CVS with new release number
   * release
   * make dist
   * double check that the tarball doesn't contain any crap
   * upload
   * go through bugs and tag them as closed in debian/changelog
   * make debian package with:
     dpkg-buildpackage -iCVS/*\|.cvsignore -rfakeroot -kFC992520
   * double check that the diff doesn't contain any crap
   * upload