* MRHOF objective function for RPL
* Om (Stanford) * Stephen (Berkeley) * Phil (Stanford)
* Om will contact John about the RPL/MRHOF implementation
Om: Today's talk is about MRHOF. Stephen, are you familiar with it?
Stephen: It's an objective function with hysteresis for RPL.
Om: Yes, it states how you compute rank from metrics.
Stephen: This is part of RPL, right? In the current way I have it set up, RPL has a routing table, and RPL does all the provisioning of the default route. All this does is change how it selects that. John is back in the loop, since he's back to working on RPL.
Om: One question: should we abstract what the metrics draft provides, a separate file. It might be good to send an implementation to the ROLL working group.
Phil: I think we should follow the specification. There's no generality: we do what the draft says we do.
Stephen: Sure, but right now the implementation uses OF0, or ETX.
Phil: Well, that doesn't meet the specification.
Stephen: What we want is, a framework, so that right now people use ETX but if there's something else people use in the future, then MRHOF will still do what the draft says. There will be a little bit of work to let you do this correctly. John is probably the right person to talk to about this.
Om: Do you think the current implementation is ready for implementing some of the ideas we've discussed?
Stephen: I think this is up to John.
Om: OK, I'll contact him.
Stephen: I'm trying to maintain the 6lowpan/RPL distinction. I do 6lowpan, John does RPL.
Om: Is there anything else I should keep in mind?
Stephen: What's important is that the interfaces down into the stack are well thought out. We want to make sure it's internally well structured. So if you see something's wrong with the interface, we should do the right thing there.