Net2WG/Notes/20080125 Meeting Notes
* CTP * Deluge * DIP
* Rodrigo, Berkeley * Om, USC * Razvan, JHU * Kaisen, UCSD * Phil, Stanford
* Om will examine new CTP trace and packet failure causes
Om: I ran a 37-hour experiment for CTP on tutornet. I haven't processed the data yet. This came from the last long experiment I ran, where we saw a 98% reliability because of a multi-hour severe dip in reliability. I'm still trying to figure out what caused this dip: at one point, a node had an 85% reliability. I think it was due to a subtree, which had only one path to the root, being disconnected.
Phil: It would be nice to see stack plots of the failure events on a per-node basis. Queue fulls, retransmit drops, duplicates, etc.
Razvan: I didn't have time to test micaz. TOSBoot modification works, but I don't know the rest.
Phil: This is not OK. You don't need exhaustive tests. But you do need a sanity check.
Razvan: I like small checkins. I don't have micazs.
Phil: Software engineering practice has shown this to be a bad idea. You should always test before a check in.
Om: OK, back to DIP.
Kaisen: I have a test script checked in. One script generates the test code: items and new items. Then you compile. When each node receives a new item, it blinks 0. When it receives all items, it turns on all LEDs. On each item reception, it sends a message to the UART.
Om: I'll test it on my desk.
Kaisen: OK, I'll test it on mirage.
Om: I've been talking to the TYMO author, and he's very interested in getting it in. Matus and 6lowpan are also gathering some steam.