- 10 Mar, 2007 28 commits
-
-
Rémi Denis-Courmont authored
-
Rémi Denis-Courmont authored
-
Rémi Denis-Courmont authored
This is not very useful at the moment, but will be needed for proper replay attack protection
-
Rémi Denis-Courmont authored
-
Filippo Carone authored
-
Rémi Denis-Courmont authored
It has to be at least 64, and it is too complicated to implement a bigger value, so it's not going to be configurable
-
Jean-Paul Saman authored
-
Jean-Paul Saman authored
Revert [19273] because these asserts are meant to catch serious programming errors. Thanks zorglub for enlightning me.
-
Rémi Denis-Courmont authored
-
Rémi Denis-Courmont authored
-
Jean-Paul Saman authored
-
Jean-Paul Saman authored
-
Jean-Paul Saman authored
-
Filippo Carone authored
-
Jean-Paul Saman authored
-
Jean-Paul Saman authored
-
Jean-Paul Saman authored
-
Jean-Paul Saman authored
-
Rémi Denis-Courmont authored
-
Jean-Paul Saman authored
-
Jean-Paul Saman authored
-
Jean-Paul Saman authored
-
Jean-Paul Saman authored
-
Rémi Denis-Courmont authored
-
Rémi Denis-Courmont authored
-
Rémi Denis-Courmont authored
Uses libgcrypt, as we have it in our deps/contrib already for GnuTLS. This could be used in both "UDP"[1] access and access output plugins, though they should really be called "RTP" instead nowadays. Done: - AES(-128) Counter Mode key derivation and RTP en-/decryption - test vectors for AES-CM - NULL cipher (with the *_UNENCRYPTED flags) - untested To do (missing mandatory features): - RTCP en-/decryption - HMAC-SHA1 authentication - replay attack protection Also to probably do: - integrate with udp access and access output plugins - integrate with RTSP server (err, I won't do that myself) - support for Transform Carrying ROC for SRTP (RFC4771) so we can use it easily for multicast streaming
-
Rémi Denis-Courmont authored
that are not plugins neither in the core
-
Pierre d'Herbemont authored
-
- 09 Mar, 2007 5 commits
-
-
Pierre d'Herbemont authored
-
Filippo Carone authored
-
Olivier Aubert authored
-
Olivier Aubert authored
-
Olivier Aubert authored
-
- 08 Mar, 2007 7 commits
-
-
Filippo Carone authored
-
Laurent Aimar authored
-
Christophe Mutricy authored
-
Rémi Denis-Courmont authored
-
Rémi Denis-Courmont authored
-
Derk-Jan Hartman authored
* use locking on the OpenGL context (using two APIs, but since agl needs the cgl headers included, should be ok i think). refs #668 refs #915 I'm not sure its fixed, but this seems like a logical way to solve it. Freaking apple with their 5 OpenGL APIs.
-
Jean-Baptiste Kempf authored
-