Re: cyradm: perl: symbol lookup error?

2015-09-16 Thread Robert Norris
On Thu, 17 Sep 2015, at 02:33 AM, Patrick Goetz wrote:
> pgoetz@www:~$ cyradm --user administrator localhost perl: symbol
> lookup error: /usr/lib/perl5/site_perl/auto/Cyrus/IMAP/IMAP.so:
> undefined symbol: Perl_xs_apiversion_bootcheck
>
>
> I'm running Arch linux, which aggressively updates software packages.
> Apparently some Perl upgrade broke cyradm?

Perl modules aren't binary-compatible across major releases (the second
number, 22 in your case - Perl versioning is a little odd). If you
recompile Cyrus::IMAP against the new Perl it should all just come back
to life. I haven't done that in isolation before but perl/imap/README
looks correct from what I know of Cyrus and Perl. Give it a try.

Rob N.

Cyrus Home Page: http://www.cyrusimap.org/
List Archives/Info: http://lists.andrew.cmu.edu/pipermail/info-cyrus/
To Unsubscribe:
https://lists.andrew.cmu.edu/mailman/listinfo/info-cyrus

Re: LAS shoutout for FastMail

2015-01-11 Thread Robert Norris

Thanks for the heads up! Its pretty exciting to see all the interest
in JMAP :)

Cheers, Rob N.


On Mon, 12 Jan 2015, at 10:02 AM, Patrick Goetz wrote:
 The FastMail guys got a nice shout out on the most current Linux
 Action Show:

 https://www.youtube.com/watch?v=TFfvpiUMYsI

 The FastMail segment starts at around 21:30 into the video.


 
 Cyrus Home Page: http://www.cyrusimap.org/ List Archives/Info:
 http://lists.andrew.cmu.edu/pipermail/info-cyrus/ To Unsubscribe:
 https://lists.andrew.cmu.edu/mailman/listinfo/info-cyrus


Cyrus Home Page: http://www.cyrusimap.org/
List Archives/Info: http://lists.andrew.cmu.edu/pipermail/info-cyrus/
To Unsubscribe:
https://lists.andrew.cmu.edu/mailman/listinfo/info-cyrus

Re: duplicatesuppression -- why?

2015-01-10 Thread Robert Norris

On Sat, 10 Jan 2015, at 09:33 PM, Patrick Goetz wrote:
 How often does it happen that the same message is being
 delivered twice?

Quite often when mailing lists are involved. You might get a reply
addressed to you directly and another one via the list. Or if you're
sending to a list, you might get a copy from the list when you already
have the message in your sent folder. In both those cases it can be
useful for the second copy to be dropped.

The major problem with it in my experience is that you might actually
prefer the copy of the message that came through the list. For me that's
usually because I wanted DKIM headers or similar. Most of the time
though its a nicer experience for users, which is why we quite
deliberately have it enabled at FastMail.

Cheers, Rob N.

Cyrus Home Page: http://www.cyrusimap.org/
List Archives/Info: http://lists.andrew.cmu.edu/pipermail/info-cyrus/
To Unsubscribe:
https://lists.andrew.cmu.edu/mailman/listinfo/info-cyrus

Re: idled vs. notifyd ?

2014-12-15 Thread Robert Norris

On Tue, 16 Dec 2014, at 09:09 AM, Patrick Goetz wrote:
 I hope I got that right. The SMS/XMPP thing could be useful if it were
 documented. I'm not sure I'm seeing the utility of the sievenotifier
 setting (send an email to let you know an email has been sent?).

Its useful if you want to notify another address that a message arrived.
FastMail's notifyd allows extra options like summarising the email.
Imagine sending it through to bridging service like IFTTT.

But yes, the SMS and IM options are far more useful.

 I still have no idea how the Fastmail notifyd

This part is the same, just with more options. The various notify
options we have are documented here:

https://www.fastmail.com/help/technical/sieve.html[1]

 works or how this is integrated with calendar functionality.

Calendar notifications are generated through mboxevents, which still
arrive via the notifyd channel but with a slightly different format.
The event has all the information about the calendar item in it, and
we use that to create an email or generate an eventsource push to the
web client.

https://github.com/brong/cyrus-imapd/blob/fastmail/imap/mboxevent.c#L146
https://github.com/brong/cyrus-imapd/blob/fastmail/imap/mboxevent.c#L858

For the alarms themselves, we keep a global database of the next alarm
for each item, which gets updated when a calendar item is saved. There's
a separate daemon (calalarmd) which periodically processes this database
and sends the mboxevents.

https://github.com/brong/cyrus-imapd/blob/fastmail/imap/caldav_alarm.c
https://github.com/brong/cyrus-imapd/blob/fastmail/imap/calalarmd.c

mboxevents are in Cyrus master (unsure about 2.4, I've never actually
looked at it) and of course calendar support is there. The alarm support
is (currently) specific to the FastMail branch. And yes, this is all
woefully undocumented.

I am being a little vague about all of this because I'm not quite sure
if you're just curious or seriously looking to hack on or deploy this.
If you ask specific questions we'll try to help out :)

Cheers, Rob N.


Links:

  1. https://beta.fastmail.com/help/technical/sieve.html?u=220c412f

Cyrus Home Page: http://www.cyrusimap.org/
List Archives/Info: http://lists.andrew.cmu.edu/pipermail/info-cyrus/
To Unsubscribe:
https://lists.andrew.cmu.edu/mailman/listinfo/info-cyrus

Re: idled vs. notifyd ?

2014-12-12 Thread Robert Norris

On Sat, 13 Dec 2014, at 08:37 AM, Patrick Goetz wrote:
 What calendar is it that is using notifyd like this?

The notifyd described in that blog post is not the stock Cyrus notifyd,
but our own custom daemon that speaks the notifyd protocol.

Cheers, Rob N.


Cyrus Home Page: http://www.cyrusimap.org/
List Archives/Info: http://lists.andrew.cmu.edu/pipermail/info-cyrus/
To Unsubscribe:
https://lists.andrew.cmu.edu/mailman/listinfo/info-cyrus

Re: idled vs. notifyd ?

2014-12-12 Thread Robert Norris

On Sat, 13 Dec 2014, at 11:06 AM, Bron Gondwana wrote:
 That said, I'm very tempted to push at least calalarmd back into
 cyrus-master for 2.5. You don't need to enable it if you don't
 want it.

I think that's the right thing to do. Alarms are pretty important and
they're hard to implement outside of Cyrus.

 I've already pushed the dlist notify protocol back to master, since
 it's so much more robust.

Yeah, and much less fiddly.

I'm sort of thinking I might write a little demo notifyd/pusher-style
thing (not included in Cyrus) just to show how easy it is to do some of
this stuff. Maybe Christmas hacking :)

Rob N.

Cyrus Home Page: http://www.cyrusimap.org/
List Archives/Info: http://lists.andrew.cmu.edu/pipermail/info-cyrus/
To Unsubscribe:
https://lists.andrew.cmu.edu/mailman/listinfo/info-cyrus

Re: Heartbleed warning - Cyrus admin password leak!

2014-04-12 Thread Robert Norris
On Sun, Apr 13, 2014, at 12:55 PM, Bron Gondwana wrote:
 Finally, as Ken mentioned, if you have an SSL-enabled Cyrus listening
 to the internet, you admin password may have been stolen already.
 Upgrading OpenSSL won't stop future login attempts with that stolen
 password.

Your private key may also have been stolen. You'll need to regenerate
your private key and certificate (or get a new from from your CA), and
get the old one revoked.

Cheers,
Rob N.

Cyrus Home Page: http://www.cyrusimap.org/
List Archives/Info: http://lists.andrew.cmu.edu/pipermail/info-cyrus/
To Unsubscribe:
https://lists.andrew.cmu.edu/mailman/listinfo/info-cyrus


Re: linker error..

2013-11-26 Thread Robert Norris
On Tue, Nov 26, 2013, at 09:56 PM, Matthijs Möhlmann wrote:
 When I try to build the above I'll get:
 matthijs@hammer:~/t$ g++ -std=c++11 -o main -lcyrus -lcyrus_min main.cpp
 /tmp/ccqJI9xZ.o: In function `main':
 main.cpp:(.text+0x26): undefined reference to
 `imclient_connect(imclient**, char const*, char const*, sasl_callback*)'
 collect2: error: ld returned 1 exit status

It looks like the Cyrus headers don't force C linkage when being
compiled for C++. Try:

  extern C {
  #include cyrus/imclient.h
  }

Cheers,
Rob N.

Cyrus Home Page: http://www.cyrusimap.org/
List Archives/Info: http://lists.andrew.cmu.edu/pipermail/info-cyrus/
To Unsubscribe:
https://lists.andrew.cmu.edu/mailman/listinfo/info-cyrus


Re: CalDAV and virtual domains not working

2013-07-22 Thread Robert Norris
On Tue, Jul 23, 2013, at 05:58 AM, Ken Murchison wrote:
 I think I prefer

 /dav/calendars/user/robn%40fastmail.fm/Default/

 as long as clients support it.

Yeah, we like this one. Its much clearer for the user.

Can't think of any reason that clients wouldn't support it - URI
escaping is hardly new. Unless calendar clients are especially
braindead? :)

Cheers,
Rob N.

Cyrus Home Page: http://www.cyrusimap.org/
List Archives/Info: http://lists.andrew.cmu.edu/pipermail/info-cyrus/
To Unsubscribe:
https://lists.andrew.cmu.edu/mailman/listinfo/info-cyrus


CalDAV and virtual domains not working

2013-07-21 Thread Robert Norris
It looks like the CalDAV stuff is not doing the right thing with virtual
domains (beta6 and git e415f906)

  2013-07-21T20:44:55.580301-04:00 calendar1 calendar/http[30755]: login: 
vpn94.mail.srv.osa [10.203.0.94] r...@fastmail.fm Basic User logged in
  2013-07-21T20:44:55.580929-04:00 calendar1 calendar/http[30755]: 
mlookup(user.robn@fastmail^fm.#calendars.Default) failed: Mailbox does not exist
  2013-07-21T20:44:55.581141-04:00 calendar1 calendar/http[30755]: 
vpn94.mail.srv.osa [10.203.0.94] as r...@fastmail.fm with Mozilla/5.0 (X11; 
Linux x86_64; rv:17.0) Gecko/20130630 Icedove/17.0.7 Lightning/1.9b1; 
PROPFIND /dav/calendars/user/robn%40fastmail.fm/Default/ HTTP/1.1 (depth=0) 
= 404 Not Found (error=Mailbox does not exist)

Mailbox structure via IMAP is:

  . LIST  *
  * LIST (\HasChildren) . user.r...@fastmail.fm
  * LIST (\HasChildren) . user.robn.#calend...@fastmail.fm
  * LIST (\HasNoChildren) . user.robn.#calendars.defa...@fastmail.fm
  * LIST (\HasNoChildren) . user.robn.#calendars.in...@fastmail.fm
  * LIST (\HasNoChildren) . user.robn.#calendars.out...@fastmail.fm

Bron tells me (if I understood him correctly) that the domain splitting
is broken, and it should be producing eg
fastmail.fm!user.robn.#calendars.Default instead.

Cheers,
Rob N.

Cyrus Home Page: http://www.cyrusimap.org/
List Archives/Info: http://lists.andrew.cmu.edu/pipermail/info-cyrus/
To Unsubscribe:
https://lists.andrew.cmu.edu/mailman/listinfo/info-cyrus