[Freedombox-discuss] Suitability of Globalscale as partner?
On Tue, 05 Apr 2011 18:11:41 -0700, John Gilmore <gnu at toad.com> wrote:
...
> This company doesn't look like the sort we want to base our software
> on.
They also have a history of producing things that I would be nervous
about leaving plugged in my home (I'd probably be OK with one in my
garage, in a sand filled box, with a fire suppressant system overhead
;-)
Initially I thought that the plug computer was a good idea, but if
you're skimping on the power supply components, it just wastes power,
and dumps the resulting heat into the CPU, which strikes me as suboptimal.
Given that Eben seems to be talking about wall-warts, and that these
sheva-plugs from globalscale seem to be what he's talking about, I think
we need to address this issue, since I think these particular
incarnations of wall-wart provide a rather shaky foundation for the
project.
Given that we have ambitions to have these in every home, even if they
have only have very slightly dodgy power supplies we're going to be
causing house fires just due to sheer numbers.
> Either they're undercapitalized (e.g. they're taking our payments
> and using the money to buy the parts to build the products - they have
> no inventory) or they don't care much about supporting small
> customers.
Check out their (total lack of) response to queries about u-boot support
of alternative boot media for the OpenRD to see how much they care about
supporting small customers.
> We could spend years making great software that runs on
> these boxes, and yet the ordinary joes who want to run it will get
> stuck for weeks dealing with a flaky vendor, give up on us, and lose
> their freedom.
Absolutely -- I think that a worthwhile expenditure of some of the FBF
funds would be developing a decent reference design, as Open Hardware,
where some bean counter has not decided to save 20 cents on the PSU
components, and where all the components are selected firstly on the
basis that it's provided by someone that's willing to support Free
Software drivers, and only then on performance, so we don't end up with
the Marvel wifi on openmoko situation, where there are bugs that will
never be fixed.
I know for example that simtek would be willing to do the layout and
publish the result as Open Hardware for what seems like a reasonable
price to me.
Once we have the design, we could get a small run made for us
enthusiasts, and meanwhile encourage other manufacturers to take the
design and run with it.
Given the wall wart problems mentioned, I've come to the conclusion that
the right thing to do is design something that is capable of running on a
mobile phone USB charger -- these are so cheap as to be almost free, and
many people are going to have spares laying around. They ahd the
advantage of massive economy of scale, and would keep the heat away
Reply to: