Re: [Telepathy] telepathy-sofiasip parameters

2011-01-31 Thread David Laban
 Hi!
 
 I'm working on a telepathy-accounts-kcm plugin for telepathy-sofiasip and
 the in-code documentation of the provided configuration parameters are not
 sufficient for me to be able to produce a sane configuration UI. Therefore
 some questions concerning telepathy-sofiasip's parameters:
 
 discover-stun, stun-server, stun-port:
 In the code i found that discover-stun is ignored if stun-server is set,
 therefore if a user sets discover-stun to TRUE stun-server needs to be
 empty to make this configuration work?
 
I didn't know this is how it worked. Just have a ticky-box for discover, and 
otherwise let the user fill in stun server and port if this is how it seems to 
work? (it might be sensible to change the stun server configuration to be the 
same as gabble's once we do ICE)
 local-ip-address / local-port:
 what are these used for, are they overriden by anything, does
 discover-binding affect this?
 
I think this just changes some headers in SIP packets. Most sip servers that 
understand NATs shouldn't need this to be changed.

 proxy-host / port / transport:
 does proxy-host need to have the format sip:FQDN/IP?
 is transport really related to the proxy?
 
for people behind firewalls, you can have a setup like:

  FIREWALL
|
sofiasip - outbound proxy -+- registrar (login server)
| +- other SIP server - other user

 extra-auth-user / extra-auth-password:
 what are extra authentication challenges?
 
in the above diagram, other SIP server may also require authentication. 
Extra auth user/pass are used for this. The primary auth credentials are used 
if these don't exist. We are currently limited to one extra set of sip 
credentials. SIP authentication challenges are actually SASL, so we could in 
theory do away with all this nonsense and pop up a SASL channel whenever 
someone asks us to authenticate.
 general:
 since i do not exactly know how sip NAT traversal works, i guess that stun,
 proxy and discover-binding (which seems to make sofiasip use natify and
 rport) exclude another?
 Are Natify and rport being used if there is a STUN server?
 
Not *completely* sure either. I think that the discover-binding stuff doesn't 
actually affect NAT traversal for VoIP calls because the information doesn't 
get communicated to farsight. Again: I am thinking of ripping up some of this 
code when we port to Call and start using ICE, so you can probably hide most 
of these options.
 
 Thx,
 
 Florian Reinhard
 ___
 telepathy mailing list
 telepathy@lists.freedesktop.org
 http://lists.freedesktop.org/mailman/listinfo/telepathy
___
telepathy mailing list
telepathy@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/telepathy


Re: [Telepathy] telepathy-sofiasip parameters

2011-01-31 Thread mikhail.zabaluev
Hi,

 -Original Message-
 From: ext Florian Reinhard [mailto:florian.reinh...@googlemail.com]
 Sent: Saturday, January 29, 2011 5:57 PM
 To: telepathy@lists.freedesktop.org
 Cc: Zabaluev Mikhail (Nokia-MS/Helsinki)
 Subject: telepathy-sofiasip parameters
 
 I'm working on a telepathy-accounts-kcm plugin for telepathy-sofiasip
 and the
 in-code documentation of the provided configuration parameters are not
 sufficient for me to be able to produce a sane configuration UI.
 Therefore
 some questions concerning telepathy-sofiasip's parameters:
 
 discover-stun, stun-server, stun-port:
 In the code i found that discover-stun is ignored if stun-server is
 set,
 therefore if a user sets discover-stun to TRUE stun-server needs to be
 empty
 to make this configuration work?

Yes, as a parameter passed to the CM it should be empty.
I think now that it is inconvenient to a plugin implementation that would not 
want to lose the server name value even if discovery is enabled. Can you file a 
bug about it? Do you want this fixed in telepathy-sofiasip 0.6?

 local-ip-address / local-port:
 what are these used for, are they overriden by anything, does discover-
 binding
 affect this?

These affect the socket binding of the SIP UA. It's a way to ensure that 
Sofia-SIP uses a particular local IP binding on multi-homed hosts, something 
that it does not do in a satisfactory way automatically.
These should not be exposed in account configuration, and are provided for 
various automatic tweaks.

 proxy-host / port / transport:
 does proxy-host need to have the format sip:FQDN/IP?

No, it's supposed to be a FQDN or an IP address only.

 is transport really related to the proxy?

This will be the transport used to contact the proxy, so yes.
I think it also goes into the pre-set Route header for initial requests, if 
loose-routing is enabled. The route URI is mainly composed from the three 
parameters above.

 extra-auth-user / extra-auth-password:
 what are extra authentication challenges?

This is almost obsolete now that we have authentication channels.
We only need to implement digest authentication on demand in telepathy-sofiasip 
to make these parameters unnecessary.
Currently, these values are used in response to authentication challenges not 
belonging to the registration domain (as discovered from the REGISTER 
authentication).

 general:
 since i do not exactly know how sip NAT traversal works, i guess that
 stun,
 proxy and discover-binding (which seems to make sofiasip use natify and
 rport)
 exclude another?

No, STUN settings are for media streams only. The UAC implementation in 
sofia-sip currently does not use STUN for NAT traversal; it relies on the rport 
RFC and has somewhat rusty and partial support for what was the SIP outbound 
draft (now RFC 5626). discover-binding enables registration binding updates 
based on external IP address reciprocated from the proxy using rport.
Proxy settings and discover-binding are orthogonal.

 Are Natify and rport being used if there is a STUN server?

Yes.

Looks like I need to spend some time documenting all these properties better.

Hope this helps,
  Mikhail
___
telepathy mailing list
telepathy@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/telepathy


Re: [Telepathy] telepathy-sofiasip parameters

2011-01-31 Thread Kai.Vehmanen
Hi,

On 31 Jan 2011, David Laban wrote:
 since i do not exactly know how sip NAT traversal works, i guess that
 stun, proxy and discover-binding (which seems to make sofiasip use natify
 and rport) exclude another?
 Are Natify and rport being used if there is a STUN server?

 Not *completely* sure either. I think that the discover-binding stuff
 doesn't actually affect NAT traversal for VoIP calls because the information
 doesn't get communicated to farsight. Again: I am thinking of ripping up some
 of this code when we port to Call and start using ICE, so you can probably 
 hide
 most of these options.

NAT traversal for media (RTP) and signaling (SIP) are two different
things, so you cannot really get rid of anything on signaling side
when you start using ICE (which is for media traversal).

On signaling side, the official solution to this problem is sip-outbound,
but sofia-sip does not yet fully support it (albeit partial support
is already present and has been for a while):
http://tools.ietf.org/html/rfc5626

Both STUN and binding-discovery are basicly ways to do signaling NAT 
traversal with proxies that do not provide any support for client NAT 
traversal (they assume client can provide a publically routable IP:port 
contact on its own).

In real-life, there are plenty of proxies that fall in between: they
don't support full outbound yet, but they still help clients with 
NAT traversal (most important bit is that they ignore the SIP contact 
field and instead send signaling to the same port client's REGISTER came
from). Even with these proxies, the keepalive options of telepathy-sofiasip
might still be useful.

___
telepathy mailing list
telepathy@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/telepathy


[Telepathy] ANNOUNCE: telepathy-qt4-yell 0.1.0

2011-01-31 Thread Andre Moreira Magalhaes
The beginning release.

TelepathyQt4Yell is an extension library for TelepathyQt4. It adds
support for some DRAFT interfaces that could not land in TelepathyQt4
itself due to API/ABI stability rules as well as some not-yet-final
utility classes.
Once the interfaces are final and the implementation is ready, the code
will move to TelepathyQt4 itself and removed from here.

Tarball:
http://telepathy.freedesktop.org/releases/telepathy-qt4-yell/telepathy-qt4-yell-0.1.0.tar.gz
 

Signature:
http://telepathy.freedesktop.org/releases/telepathy-qt4-yell/telepathy-qt4-yell-0.1.0.tar.gz.asc
 

The latest reviewed code is always available from:
git://git.collabora.co.uk/git/telepathy-qt4-yell.git
http://git.collabora.co.uk/?p=telepathy-qt4-yell.git (gitweb)

The first release of telepathy-qt4-yell features:

 * High-level API and auto-generated client classes for Call.DRAFT.
 * Models support.

Regards,
Andre

___
telepathy mailing list
telepathy@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/telepathy


[Telepathy] ANNOUNCE: folks 0.3.4

2011-01-31 Thread Travis Reitter
libfolks 0.3.4 is now available for download from:

http://ftp.gnome.org/pub/GNOME/sources/folks/0.3/

cd35e7d0ecd57cea505e661c69140690587c427b846183f0e18855b15feee1a9
folks-0.3.4.tar.bz2

68aa5fc6e665e511a2421167012e4b94ee8c51fff81ea64b60e126bca8cfa984
folks-0.3.4.tar.gz

libfolks 0.3.4 — I get by with a little help from my friends

Libfolks pulls together contacts from any number of accounts supported
by the libfolks backends. This release includes a Telepathy backend
complete enough to power Empathy and a key file backend to allow contact
linking. We would like to include additional backends
(especially evolution-data-server), so let us know if we can help!

libfolks is not yet API or ABI stable, so please proceed with caution.
See the NEWS file for details on every break we make. In future
releases, we will have example programs that will be kept up-to-date
with the latest API as well.

For more information, see the Folks wiki page:
http://telepathy.freedesktop.org/wiki/Folks

What's new?
===

Major changes:
* Add folks-inspect tool
* The use of the new LinkedHashSet type in the API may make libgee a
  requirement  for some clients (to take full advantage of its class
  lineage).

API changes:
* New type: LinkedHashSet, similar to Gee.HashSet, but ordered.
* IMable.im_addresses is now a mapping of string to LinkedHashSet.
* Rename the HasAvatar interface to AvatarOwner
* Rename the HasPresence interface to PresenceOwner

Bugs fixed:
* Bug 637240 — libfolks-telepathy.so exports private symbols
* Bug 638311 — Add a HACKING file that outlines development policies and
  coding style
* Bug 629083 — Review coding conventions in folks
* Bug 629075 — Add folks command line application
* Bug 638609 — libfolks hard-codes backend names for debugging
* Bug 639195 — Print stack traces for failed tests to improve remote
  debugging
* Bug 639113 — Add static aggregation tests
* Bug 639742 — Logger service unavailable in make check
* Bug 640213 — Add tests for LinkedHashSet
* Bug 627397 — Use better interface names

Dependencies

libfolks and depend upon core GNOME libraries (GLib, GIO, etc.), as well
as the recent releases of Vala and gobject-introspection.

The Telepathy backend also requires a recent release of
telepathy-glib, built to include its Vala bindings.

Contributing/Contact

For any questions, comments, feature proposals, etc., please refer to
the Telepathy mailing list or #telepathy on Freenode.
For bug reports, please file them with the GNOME Bugzilla module folks

___
telepathy mailing list
telepathy@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/telepathy