Disable codecs for iOS that does not support videotoolbox.

Change-Id: I3949d4ad262ec4393e954a8e279b6f316861d883

Change-Id: Ia80b1f8c8bb43432739d409557b2995cc66a575f

Change-Id: I4ed20ac1e40df5261c5b0a4cb8dd74fdb97ab17d

Change-Id: I2a590a5e4a63279582ea605b7fbf10d37ec4c571

Change-Id: Ie568064fdeb4f40a77fa7a6f3e0946c10df828ef

Since on iOS app have different path each time when starts, ringtone
paths also must be updated. But when playing ringtone we all time could
have current application direction, so we could just save relative
constant path.

Change-Id: I738e5bd26d036b11201a8b2a83bb22126d3139e4

Change-Id: I3a7cbf15795d1f1f015328054649ab354d9c8716

Change-Id: I44a55c88d01a0df0695814a037d74d4425c09c65

Change-Id: Iad0b1f93cfa41ab7547157604978a31fdd71e85f

Philippe Gorley authored 5 years ago and Sébastien Blin committed 5 years ago

MediaRecorder only calls the MediaEncoder with an AVFrame*, not with our
VideoFrame/AudioFrame wrappers, so the audio stream was never
initialized.

Gitlab: #95
Change-Id: Ie9c85eb9334444f2b58a4355e8f6db9d2a14610f
Reviewed-by: Sébastien Blin <sebastien.blin@savoirfairelinux.com>

Change-Id: I055cc4bd496966b1f1306fc7c7dfd23f9ef6fe25

Change-Id: I6cd7e28ed62581c5635a32966b46171441de42c2

Change-Id: I919aa2607b1d0dcc33b7b02f90b97b1b8fd1ef98

Change-Id: Ic959e2db4a484bde5c19fa1bec7a997cb332f571

Change-Id: I3bd3b4918cf2a7a9a37180d4898e291d07565988

Change-Id: I993003f8d448a1181ecdd167df10bedc2048b76e

Sébastien Blin authored 5 years ago and Adrien Béraud committed 5 years ago

Change-Id: I992bee823098e100a67da3bcb28c67722370d618
Gitlab: #111

Mohamed Fenjiro authored 5 years ago and Sébastien Blin committed 5 years ago

Change-Id: I57d3efa477490196e6e935d00a6e98f15d970714

Change-Id: Ia6a43fe3650762b0b8cf1185aaac6cf582b55b8b

avoid deadlock by locking mutex on next loop like
other pjsip callbacks

Change-Id: I6e227fc961c41d35c041c907eb2bc73dbb373119

Change-Id: Iecdceb428e0c174883c94754444c0046606e4bcb

Change-Id: I9b440d96fc34e9442b7856e428190cc8110f6bfa

Eden Abitbol authored 5 years ago and Adrien Béraud committed 5 years ago

Part of the discovery event handling for libupnp was to delete
all port mappings associated with the local ip address of the
application. I can only assume this was done to close pre-
existing ports that were not properly closed when the application
terminated. The problem with this logic is that since
advertisements and discovery events are treated with the same
switch case (fall through), the application was trying to close
all the ports on the internet gateway everytime it got an
advertisement. However, the application would then try to reopen
the ports every time after closing them. And this would happen
every five to ten seconds (i.e. whenever the application would
catch and advertisement or discovery event from the internet
gateway.)

To fix this quickly, I modified the way the event handling treats
discovery and advertisement events. Instead of having a list based
on the URL of the internet gateway, I use a list of it's unique
service ID. That way, as soon as the event occurs, we check if
we don't already have this service ID in our list. If we don't,
then we proceed as usual. If we do, we exit the event handling
since it's already been processed.

The advertisement bye bye event has also been implemented. When
this event occurs, the corresponding internet gateway devices are
deleted from the lists.

Gitlab: #96

Change-Id: Idd8023eba319b431b3a9328ebe648e75d61b1dc8

Adrien Béraud authored 5 years ago and Sébastien Blin committed 5 years ago

Change-Id: I31f6aa38b7773bc0ecbdeb4ba2ae556f3fd53cc3

Now that RFC6062 is merged upstream, we can remove the patch from
our stack. The API changed a bit, so this patch also updates
turn_transport.cpp to follow changes

Change-Id: If6e0bae8280d586b2e5fcb0babe83df8127789b6

This also solves calls via TURN. In fact the redefinition in
rules.mak leads to a difference between the internal offsets
in pjproject and the daemon. So, structures passed to pjsip are
incorrect.

Change-Id: I50261781511b11cffd9441e825d8edb09256f3e0
Gitlab: #105

Adrien Béraud authored 5 years ago and Sébastien Blin committed 5 years ago

Change-Id: I86be545fb21b637accade1867ab9ff26b652dd8b
Reviewed-by: Sébastien Blin <sebastien.blin@savoirfairelinux.com>

Change-Id: If4c13f20b1bbf2e0cf9c6e78ffe4d68b99e98842

Philippe Gorley authored 5 years ago and Sébastien Blin committed 5 years ago

Change-Id: Iaac2793ce2a6e0d32027afe168b1511ef0575016
Reviewed-by: Philippe Gorley <philippe.gorley@savoirfairelinux.com>

Mohamed Fenjiro authored 5 years ago and Adrien Béraud committed 5 years ago

Change-Id: I719a83badfb811ab9bb11dc01d29ce9363829573

Change-Id: I6c3256ebe4192d02fcc111ff6a7df92456c3d2f2

Adrien Béraud authored 5 years ago and Sébastien Blin committed 5 years ago

Change-Id: I0057179d4b02e19a7d5663ba0e1d3e2ab1588cb7

Adrien Béraud authored 5 years ago and Sébastien Blin committed 5 years ago

Change-Id: Ibb08cc5363e6e5964856b30f6a899f5b25e56341

Adrien Béraud authored 5 years ago and Sébastien Blin committed 5 years ago

Change-Id: If34adb4833dff7ea71175bb7f465f48cb3758d55

Change-Id: I2bfbde86083829d553a1f79e1c91aed4b2bedd6f

Sébastien Blin authored 5 years ago and Adrien Béraud committed 5 years ago

Because the signature is now fixed and sign(Name) != sign(name),
the daemon should sign the name in lowercase because it will
try to register the name in lowercase.

Change-Id: If5f786f241fedb83d65015ac7edf0cacd66f4d2d

Change-Id: If65e96d9ae67631deda902240a828b084ec8611a

Change-Id: Ibb7ed1c76dafc7368655b0751f94a4b0df6a531d

Adrien Béraud authored 5 years ago and Sébastien Blin committed 5 years ago

Change-Id: I8ef7df59cd972868fda4e95dc0cc36ba123ffa44

Adrien Béraud authored 5 years ago and Sébastien Blin committed 5 years ago

Change-Id: Iff27c0ca7efedfd54bb5b1763fd2178819128172
Reviewed-by: Sébastien Blin <sebastien.blin@savoirfairelinux.com>