On Thu, Mar 26, 2015 at 1:16 PM, Juliusz Chroboczek <[email protected]> wrote: >> I would like to know what happens if Dave Taht installs your Babel >> implementation on half of his base stations and keeps Juliusz' on the >> other half. > > Markus' implementation doesn't currently do any attempt at link quality > estimation, so any pybabel-pybabel link will be treated as if it didn't > suffer any loss, which will cause long and highly lossy links to be > preferred. I'm pretty sure that would break Dave's network, we'd need to > test in a somewhat less hostile environment.
Heh. Yes, I have one intentionally hostile environment... but... I have multiple production and lab networks, and python is at least partially supported by cerowrt, and fully supported by openwrt, so someone else could test it easily. In my case several of my more hefty routers run ubuntu or are debian based like the ubnt edgerouter - and furthermore it's quite routine to just add a roaming laptop running linux or OSx to also be a participating router... so I can certainly test it pretty easily in one or more of my own labs, as could anyone else willing to try it. That said, I'm more tempted to burn a week rewriting the python into rust or go, now that I know how easy it is. (thx, markus, for completely blowing my mind!) Part of that effort would be tho to get a better kernel netlink implementation and low level network API interface than either language has already (someone update me if there is something for those that can do sendmmsg and recvmmsg?), and to explore any parallization opportunities that exist for finding the best route using either of those languages' native primitives for co-(go)routines with a large (10000s) number of routes. Does python not have a decent netlink interface already? > As to wired nets, the only limitation I can see right now is that pybabel > doesn't disable the Linux kernel's uRPF filter, so unless you remember to > do it manually, you're going to get transient blackholes. I'll hopefully > be able to tell you more once I've finished reviewing the code and done > some testing. > > By the way -- what Markus has written is some compact, elegant and > comprehensible code, and the comments include pointers to RFC 6126. If > you grok Python, and if you're the kind of person who understands things > better by looking at code, his is the implementation to look at. I don't grok python very well, but it is mildly easier to read than the c, yes. lua, anyone? > -- Juliusz > > _______________________________________________ > homenet mailing list > [email protected] > https://www.ietf.org/mailman/listinfo/homenet -- Dave Täht Let's make wifi fast, less jittery and reliable again! https://plus.google.com/u/0/107942175615993706558/posts/TVX3o84jjmb _______________________________________________ homenet mailing list [email protected] https://www.ietf.org/mailman/listinfo/homenet
