[Milkymist-devel] Milkymist as set top box

Werner Almesberger werner at almesberger.net
Fri Oct 7 04:28:58 EDT 2011

Jon Phillips wrote:
> I guess I'm asking, what would it take to make milkymist work for
> normal people to get media (audio and video) from a computer to a tv,
> with the milkymist being at the endpoint connected to a LCD panel with
> a ethernet cable.

So the "LCD panel" would be the TV ? Or is this a separate entity,
e.g., a control panel dedicated to the M1 ?

If the idea is that the M1 would generate the image shown on the TV,
you would need:

- access from the M1 to the media source. If the M1 itself acts as
  the media server, this would be implicit. Otherwise, it would
  have to support a suitable set of protocols for file/media sharing,
  e.g., NFS, SMB, UPnP, etc.

  If Internet streaming is a potential media source, it would have to
  implement this as well.

- the ability to decode and play/render the media stream in real
  time. Depending on what you're after, this may require support for
  proprietary codecs.

If the M1 itself acts as the media server, it would have to support
a reasonable set of protocols to serve content to other systems,
e.g., your laptop.

Finally, returning to your original usage scenario, you'd need some
protocol to tell it to render what's on your laptop. UPnP may have
such things already.

That's a lot of missing items. Most of the protocols would be
available with relatively little effort once Linux runs properly on
the M1. But you'd still need the codecs. I'd expect the LM32 to be
way too slow for doing much with software-only decoding. Supporting
proprietary codecs would also get you into the well-known licensing
issues. And not supporting them would make this solution interesting
to only a rather small population.

I think there are quite a lot of lower-hanging fruits to improve the
M1's value for the functionality we're marketing it for today. And I
hope we can get developers interested in helping with these.

Some things would coincide. E.g., I think having proper Linux support
will be a way out of USB driver hell and it would in turn give access
to other features, such as supporting the kind of communication
protocols I've listed above.

- Werner

More information about the discussion mailing list