Hello,
I plan to unify processing of voice RTP data in wireshark. I plan to
focus on decoding data for replaying (player, save as au etc). The
target is to be able to play multiple streams in player which requires
to mix multiple streams together.
I reviewed code and found that part of code stores temporary voice
data in memory, part in temp files. When I unify code, only one method
must be used.
Do anyone has any points to discuss about this topic?
There are my points:
Storing in memory was used by GTK code and is used by most library
methods. When RTP stream is long, it consumes memory, but provides
faster processing. If there will be a lot of long RTP streams, memory
can be depleted.
Storing on disk avoids depletion of memory, but e.g. mixing of
multiple streams is slower.
Moving in stream during replay (start at, forward, back, ...) do not
depend on method of storing data.
Any comments are welcomed.
Best regards,
Jirka Novak