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

[Dave Taht <[e-mail]>]

On Thu, Oct 22, 2015 at 4:09 AM, Douglas Otis <[e-mail]> wrote:
>
>
> On 10/21/15 5:29 PM, Dave Taht wrote:
>> On Wed, Oct 21, 2015 at 5:19 PM, Henning Rogge <[e-mail]> wrote:
>>> On Wed, Oct 21, 2015 at 5:10 PM, Dave Taht <[e-mail]> wrote:
>>>> is it up from 8?
>>>>
>>>> Dave Täht
>>>
>>> I did experiments in the CORE network emulator with shncpd... not sure
>>> if this counts.
>>>
>>> Henning Rogge
>>
>> It counts! Has the channel assignment problem been solved in any way?
>>
>> Dave Täht
>> I just lost five years of my life to making wifi better. And, now...
>> the FCC wants to make my work, illegal for people to install.
>
> Dear Dave,
>
> I share your concerns. Perhaps DD-WRT could include a canned
> library with a flexible API. Perhaps we should defined this
> API and not settle for something like NDIS. An API that
> isolates power related adjustment code to meet regulator
> requirements.  We have lived through the regulator hell in
> the recent past. It was not pleasant.

This code has existed in linux for over 10 !@#! years. It is
used by default in most systems. The regs invoked last year
and upcoming this year are massive overreach.

The API is bog-standard. The code is under a liberal licesne.

The freq database is signed, and frequently updated, for all worldwide
standards.

https://wireless.wiki.kernel.org/en/developers/regulatory

> Once specifications become clear, we also hope to do
> testing. Are there good places to find code implementing
> homenet dncp with scalable DNS-SD,

most homenet code and spec were written at the same time.

openwrt has the needed proxy available for it. How scalable,
dunno. Ask barth.

> and versions running ISIS
> with scalable DNS-SD as well?  I would assume wifi running
> ISIS also blocks port 5353 traffic at the bridge.

Still intending to stick with babel here. It is massively lighter
weight, well tested, fully deployed in mainline linuxes, and has other
features we have long liked.

Don't want to get into the ISIS vs babeld debate today. Do an apt-get
install babeld on any debian derived box or yum install or whatever
package manager you use.  Next question?

I would like to see source specific routing enter all routing
protocols and all routing daemons, actually.

>
> What is your opinion on a good router for testing?  Not
> necessarily the cheapest because performance is also
> important.  Should I look for the best DD-WRT hardware
> running linux before these units are removed from the home
> market?

I note we work primarily on openwrt. dd-wrt lags behind a lot.

We are working on getting the linksys ac1900 up to speed
but the binary blob in the wifi driver is messing up that universe.
(aside from that, pretty good hardware).

A highly anticipated new platform is this:

http://wiki.openwrt.org/toh/turris/turris

which basically uses the same chipset as the above, but
gives us options on what wifi cards to use.

On the lower end, anything with an ath9k in it is decent, but
the hardware is showing it's age. the new cake qdisc is
40% faster than the old sqm-scripts + htb + fq_codel
system but it still barely scales to a 100/20Mbit link

And the wifi on everything sucks universally, right now.

> My fear is testing may soon require a license and
> expensive custom hardware.

Well, nobody has talked about any licensing scheme for being allowed
to do wifi work...

so aside from vendors working directly in the field, and people and
institutes willing to ignore present and future rulemaking... (or
change it)...

we're going to be f**ked, here in homenet.


Dave Täht
I just lost five years of my life to making wifi better. And, now...
the FCC wants to make my work illegal for people to install.
https://www.gofundme.com/savewifi