The previous kStrict mode could generate very small packets when the encoded frame is smaller than MTU size. kAggregate will instead encapsulate whole frame into one packet if frame size is below MTU (and so will not generate too small packets), and otherwise it will separate out the first partition as in kStrict mode.
The balanced setting for kAggregate (from default of un-balanced) is also desirable, as equal size packets (for the first and remaining partition) should generally be more favorable for FEC.
Review URL: http://webrtc-codereview.appspot.com/89002
git-svn-id: http://webrtc.googlecode.com/svn/trunk@239 4adac7df-926f-26a2-2b94-8c16560cd09d
The variable _numberFirstPartition is now set in RTPSenderVideo::SendVP8.
The number of packets that contains data from the first partition
is not known until all packets have been packetized (at least all
first-partition packets). Therefore, the packetization loop in SendVP8
had to be broken up into two loops. The first loop gets all packets from
the VP8 packetizer (RtpFormatVp8) and puts them in a vector. The second
loop sends all packets from the vector to SendVideoPacket.
Review URL: http://webrtc-codereview.appspot.com/56004
git-svn-id: http://webrtc.googlecode.com/svn/trunk@231 4adac7df-926f-26a2-2b94-8c16560cd09d
