(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Net2WG/Notes/20080321 Meeting Notes
Agenda
* CTP Beacon rates
Participants
* Om, USC
* Razvan, JHU
* Phil, Stanford
Action Items
* Om will make 3 fixes to CTP
Discussion Notes
Python Support
* Razvan: I implemented an alternate Python serial support for Deluge. It's called tinyos.py. What should I do with it? Should it be an alternative support stack for serial communication?
* Phil: You should talk with the existing serial support maintainers, David Gay and Chad Metcalf. Your approach has a different API , and the current one was designed to work with mig/ncg.
CTP Beacons
* Om: If you look at the data at [[1]];, you can see what events are causing the beacon timer to reset. About 2/3 are loops, with one major routing node dominating, while 1/3 or so are new parent events.
* Phil: It is kind of weird that new parent triggers the beacons. It should only matter if the new parent is significantly further away. But even then, this only matters if the node has to route packets.
* Om: But if my new parent has a higher ETX, then I need to let people know that I have changed.
* Phil: Maybe we can just leave this to the loop detect path. That way, we only have one thing to optimize. Same with full events. These are leftovers from the failed attempts at ECN.
* Om: OK.
* Phil: It might also make sense to make the beacon timer a true trickle timer: use suppression.
* Om: So there are three changes: suppression in beacon timer so we adapt to density, do not have new parent trigger the timer, and full does not trigger the timer either.
