Audio/Video conversation

[H@t Trick]

oh ok cool, I will still get you that gstreamer/farsight error from the status log when I have a chance, but I will try tomorrow night or sunday and confirm proper behavior with the DMZ off, this is fun to actually be testing an experimental feature

[square87]

Hi.

When i give ./utils/farsight/farsight 1 2
i get:

Code:

(farsight:9216): GLib-GObject-WARNING **: specified class size for type `FsRtpSession' is smaller than the parent type's `FsSession' class size

(farsight:9216): GLib-CRITICAL **: g_once_init_leave: assertion `initialization_value != 0' failed

(farsight:9216): GLib-GObject-CRITICAL **: g_param_spec_object: assertion `g_type_is_a (object_type, G_TYPE_OBJECT)' failed

(farsight:9216): GLib-GObject-CRITICAL **: g_object_class_install_property: assertion `G_IS_PARAM_SPEC (pspec)' failed

[kakaroto]

humm.. apparently your bug (square87) is a 1 change in 100 of getting it.. it shouldn't happen all the time.. it's a bug in glib, if you update to glib 2.16, it shouldn't happen, but since it's a race condition and would rarely happen, I don't think you would get it if you retry...

[square87]

I already tried variuos time before posting...
I have libglib 2.16.3-1 (the hardy version...)
Btw i also think that's a my problem... i'll try to reinstall/compile

Thanks, byez

[mgomes]

Hi everybody,

First of all, i would like to congratulate amsn development team for this project, i use amsn since 0.95 version, and i like so much to use this software!

Well, i've read that amsn svn has the new feature of audio call, and check the farsight wiki page to use it. I'm on ubuntu hardy, and installed the following packages from ubuntu repository:

gstreamer0.10-alsa 0.10.18-3
gstreamer0.10-plugins-base 0.10.18-3
gstreamer0.10-plugins-good 0.10.7-3
gstreamer0.10-plugins-bad 0.10.6-5
gstreamer0.10-plugins-farsight 0.12.5-2ubuntu1

And compile farsight2 - 0.0.2 using ./configure --prefix=/usr

I didn't use that packages supplied at wiki, because them all are for i386 machines, and i'll compile amsn at a amd64 environment, so i guess that's could be a problem to use i386 libraries with amd64 binary.

Well, after compiling amsn (compiled amsn using: ./configure --with-tk=/usr/local/lib/ --with-tcl=/usr/local/lib/) tcl and tk 8.5.2, i executed ./utils/farsight/farsight 1 2, and got the following:

user@machine:/usr/local/src/amsn$ ./utils/farsight/farsight 1 2
Error while creating new session (0): Could not create the fsvalve element**
** ERROR:(utils/farsight/farsight.c:305):main: assertion failed: (0)
Aborted

I don't know what's happening, i read this thread entirely, and didn't see this error before, them someone could help me?

Sorry for my bad english, and thanks in advance for the help

[trv]

why did you use /usr/local/lib in the amsn's configure options? Hardy has tcl8.5 in the repositories, for 64bit systems too. For starters, you should install tcl8.5 and tk8.5 from the repositories, and configure amsn with with-tcl=/usr/lib/tcl8.5 and with-tk=/usr/lib/tk8.5

About the farsight error, i don't know if farsight2 has been tested in 64bit pcs, maybe someone with 64bit can share his experience

[kakaroto]

Hi mgomes, Thanks for the nice comments and welcome to the forums.

You never saw the error before because noone ever did what you did before :p
in the wiki, it does say that you need to install the gst-plugins-farsight version that I linked.. that's a darcs version (like svn) of gst-plugins-farsight which is between 0.12.7 and 0.12.8... you installed 0.12.5 which is pretty old and doesn't have the necessary gstreamer plugins (like the fsvalve it's complaining about).
So please, just download the gst-plugins-farsight-siren.tar.gz from the link provided in the wiki, and compile that instead, that should fix your issue.
have fun!

[mgomes]

Hi trv, thanks for your reply! i compiled tcl and tk because i would enable the antialiasing on the fonts in amsn. I didn't know if the ubuntu packages comes with this feature. At 8.4 version didn't, and in earlier versions of amsn i compiled it to enable antialiasing.

Hi Kakaroto, thanks for your reply too! Sorry, but i didn't know that gst-plugins-farsight at ubuntu repositories is an older version, but i removed it, downloaded the tar.gz at wiki, compiled and succesfull run ./utils/farsight/farsight 1 2

user@machine:/usr/local/src/amsn$ utils/farsight/farsight 1 2
LOCAL_CODEC: 96 SIREN 16000
LOCAL_CODEC: 0 PCMU 8000
LOCAL_CODEC: 8 PCMA 8000
LOCAL_CODEC: 97 MPA 90000
LOCAL_CODEC: 3 GSM 8000
LOCAL_CODECS_DONE
LOCAL_CANDIDATE: L0 1  UDP 0 127.0.0.1 7078
LOCAL_CANDIDATE: L0 2  UDP 0 127.0.0.1 7079
LOCAL_CANDIDATES_DONE

Well, after this program working, i logged in at aMSN and called a friend that uses WLM, and i could hear him, but he didn't, then i thought this problem could be that issue you mentioned about the WLM user be on a subnet, and i tested with another friend, and works fine at WLM, i tested only to sendo audio call invitation, not receiving! Unfortunately, i still didn't test with another amsn user, but when i found one, i'll test

Once more, thank for the help, and congratulations for the project!

One more question, Kakaroto, i saw that the packages you supplied at wiki are for i386 machines, if needed i could make amd64 packages to improve the support

[kakaroto]

Hi mgomes,
yes, gst-plugins-farsight in ubuntu is not the latest, but also, you needed a 'special' release of gst-plugins-farsight with the support for the Siren codec, and there's no official release with that yet.
Anyways, thanks for taking the time to read the thread.. when I say you report that sound worked one way only, I thought "oh no, do I have to repeat myself", but then you read it :p
About receiving audio call invitation, you can only receive one if you use MSNP15, so if you don't, you can't receive it anyway, sending or receiving will work the same way, so it won't make much of a difference to test one or the other...
Yes, the packages are i386, I don't think we need amd64 unless someone requests it... anyways it's svn, an experimental feature, so they should get the right packages (by the distro) once we release amsn 0.98.. in the meantime, it was just to help more people try it faster...
thanks anyway!

p.s: if you still do it, you can send me the packages.

[H@t Trick]

Well I finally go the chance to test with the DMZ disabled and well funny, no combination seems to give any audio now, re-enable the DMZ, no audio still, this is getting frustrating now when it was excting before, but hopefully tomorrow I can retest and get you some status logs and PM them to you.

[Yannick]

Hello,

I'm out of topic, so please forgive me.

Reading the page 9 I saw some protocol logs used by WLM.

It seems to be SIP.

kakaroto, do you know if it is real SIP compliant or some custom variant? Is there documentation available somewhere about it if it is custom?

Best regards,
Yannick

[kakaroto]

Hi Yannick.. I think I pretty much answered that question in my previous posts, I've given very detailed description of what happens...  so if you need any technical info, just read the whole thread (it's a lot I know.. I wrote most of it :p)
anyways, to summarize, yes, it's 100% SIP compliant, nothing is different, everything is as is.. (I also was able to make it work using telepathy-sofiasip..) note that the abbreviated header fields ("t" = "To", "v" = Via", "m" = "Contact", etc...) are also SIP compliant (they are called "Compact headers" : RFC 3261).
A few notable differences from 'standard' SIP (but not non-rfc compliant afaik) is that the From field needs a parameter "epid" randomly generated.. and it all goes through TLS (BUT the sip URI is 'sip:" not "sips:") and the authentication used is Basic with a fixed username and the password being the MessengerSecure token received by SSO.
as I said before.. SIP/SDP/RTP/ICE/STUN/TURN/etc.. all those protocols used by this feature are 100% RFC compliant (apart from the stun message-integrity which appends some null characters before doing the hmac.. but that's just an authentification issue to make sure not anyone uses their servers.. but it still uses the same md5+hmac algorithm for authentication as described in the RFC... it's just that the message being checksummed has 60 null bytes appended to it).. note that ICE is a bit different, but that's normal (even gtalk has a different ICE implementation), that's because ICE is not yet an RFC, it's still a draft, gtalk and MSN both implement the 6th draft of ICE (there are 19 so far), and they each have some differences from the draft since the draft doesn't work perfectly. WLM changes are mainly (afaik) about the username/message-integrity authentication parameters.

[Yannick]

Hello,

Thank you kakaroto. I'm part of the Ekiga team (well, I'm not a coder myself). Damien Sandras is interested It seems, after this "discovery" for us, compatibility with WLM is on his TODO. (using SIP)

Best regards,
Keep up the good work
Yannick

[kakaroto]

hi Yannick,
I'm glad this discovery interests the ekiga team! I just hope you won't release your compatible version before ours :p If you need help porting Ekiga to use farsight, I'd gladly help (ok, just teasing :p).
Anyways, you will need SIP + RTP + ICE, and probably TURN.. I didn't release my RE work on the ICE+TURN but I will soon (once I fix up libnice which will need refactoring) implement that with farsight too..
If you need help, don't hesitate to ping me on #amsn channel at irc.freenode.net

[Fabioamd87]

http://forums.microsoft.com/MSDN/ShowPost.aspx?PostID=833356&SiteID=1

i think they ignore us