Re: [homenet] How many people have installed the homenet code?

[Tim Coote <[e-mail]>]

Hullo

Sorry, this is a bit of a granular question, and this may not be the right place for it - I’m relying on the value of running code to the group ;-) . My background is s/w engineering, rather than networking, so I could well have made some fundamentally poor assumptions.

There was a discussion back in Oct 2015 about running hnetd (I presume) on mainstream linux distros (http://bit.ly/1XKc3Oi <http://bit.ly/1XKc3Oi>). Henning Rogge raised the question and Markus Stenberg responded that it had been tried on Debian 7. The discussion also referred to building hnetd on OpenWRT (http://bit.ly/1VDSg5P <http://bit.ly/1VDSg5P>) and (http://bit.ly/1VDSs4Q <http://bit.ly/1VDSs4Q>) based on using a larger distro and shncpd for a subset of the homenet protocols. The hnetd repo implies that it can be made to work with generic linux router firmware, which I took to mean any linux distro.

I think that this is an important issue, based on my experience of trying to build home automation products in the IoT space. Environments without carefully engineered networks can behave in unexpected ways, even with small numbers of end devices and realistic deployments will require significant software instrumentation and the ability to change the instrumentation to understand what is going. So, I believe, you need a full fat distro for such environments that’s responsible, say, for handling 6LowPAN devices, abstraction of such devices for use by other services, and managing their lifecycles.

So I tried to spin hnetd up on a fedora vm, and found it fighting the distro set-up. Maybe the implementation isn’t supposed to co-exist with other network controlling software on the same computer. What’s my best approach to getting the homenet protocols running as a proof of concept? I cannot find a ‘top down’ view of how the different components fit together - the descriptions seem to assume a depth of knowledge of openwrt that I don’t have.

tia
Tim