[Bug 770935] RTP/RTSP/TCP transport RTSP TIMEOUT should start at once when tcpi_ca_state is TCP_CA_Loss

[GStreamer (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=770935

--- Comment #16 from Dag Gullberg <opaque.se at gmail.com> ---
It is an external call, but it is a mem copy of a struct as far as I
understand. I doubt it is more expensive than the socket send. To lessen the
impact I could have a counter that made the call more seldom like

#define N 2  /* inverted frequency of getsockopt calls to check TCP_CA_Loss) */
guint cnt;

..

cnt++;

if ( cnt % N == 0 ) {
 check_tcp_ca_state_loss(watch->conn->write_socket));
}

I will try to do some timing checks with the call enabled and disabled.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.