On Sun, Jul 30, 2006 at 05:36:13PM +0200, Tino Keitel wrote:
> > I agree that the latest isync is loads better than the old one, and so
> > if things don't work maybe we just tell people in the NEWS file that
> > it's an incompatible upgrade, and that things will just break, or
> > maybe we create a new mbsync package and tell people that they have to
> > explicit upgrade to the mbsync package and be ready to deal with any
> > compatibility issues that might exist.
>
> I think that a smooth upgrade without user interaction is desireable
> here.
I agree, that would definitely be best. What is clearly unacceptable
is if things break and users aren't warned about it in advance, with a
chance to abort the upgrade if they're not ready. (I've had some
really lousy failure modes where isync's ID mapping database is
corrupted, or the IMAP server's namespace changes, and I'm travelling
overseas in a location with lousy/high ping times, and am forced to
throw away the isync replicated mailbox and synchronize thousands of
messages and hudreds of megabytes of data over a slow/high latency
network link. It can highly annoying at best, and extremely costly or
completely impossible to get new mail at worst.)
> OK. I'll set up a local imap server with several folders and a few
> thousand mails, sync with the old isync, and then try the new isync.
> What should I look for? Just test if sync (also with new/deleted mails
> on the server and in the local maildirs) works? Any wishes regarding
> the server? I think I chose courier, dovecot, or both.
>
> I hope I find the time for this next weekend.
That would be great, thanks! I Think courier is more popular, but
there are people who use dovecot too. I think the biggest problem
won't be specifically with the IMAP server, but with the format of the
records stored in the berkdb database as used by isync. So if it
works with one IMAP server it's likely that it will work with the
another IMAP server, since isync doesn't try to interpret the contents
of the UID. Still, it would be good to test with both.
Thanks!!!
- Ted
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
