[mythtv-users] Architecture Idea - Raspberry Pi, VM, XBMC
Simon Hobson
linux at thehobsons.co.uk
Sat Jan 5 20:30:15 UTC 2013
jedi wrote:
> So the idea if using a PI as a slave backend is not as strange as it
>may seem. As ARM hardware gets more commonplace, I would expect that
>stuff to get sorted out. It may even work already.
>
> Someone just has to give it a go. Build V4L on a PI and see how it goes.
I guess ARM will "sort itself out" in time. It;s easy to forget just how much we take for granted in the Intel (and AMD) world these days. Some of us can still remember when "PC Compatible" meant that it ran some version of MS DOS - not that any of the hardware was in the same place or that it could run anything other than it's own customised version of MS DOS. Lots of stuff ran CP/M - and again, the CP/M for one machine couldn't runn on another machine because all the hardware would be in the wrong places.
What little I know about ARM stuff (mostly from listing to what others tell me) is that there's no standardisation. So great, you've built a kernel, but it'll only boot on the hardware you've built it for. Try and boot it on anything else and it won't be able to find the hardware because each vendor has their own random locations for everything.
Unfortunately, I don't see that changing like it did with Intel. Back then, some magazine games reviewer (IIRC) came up with the test of "shove in the boot floppy for <some game> and see if it runs it" as a test. If it ran it then it was "compatible", if it didn't then it wasn't. Thus all the vendors had to change their architectures to put (eg) the floppy controller at a fixed place, the serial ports at fixed places, and so on. These days vendors seem to be competing to see how closed they can make stuff (even if running Android) - so there's no equivalent test to embarass vendors into "fixing" their strange hardware.
More information about the mythtv-users
mailing list