[Lightget-misc] Development ideas

[Risto S V. <rva...@cc...>]

I think modularity in hardware (power supply, main unit, LED
arrays) was a really good idea.  You can verify each module
separately as you build them, and when you experiment with some
new module, you don't have to rebuild the others. It also gives
you flexibility as you can change LED colours etc. easily.

As for future directions...

Maybe we could specify some hardware interfaces, like a dumb
parallel port interface and a more advanced serial
port interface for using programmable lightget units. 

Another option would be to specify the interface for device
drivers. The hardware units could be very different, but each
would have a dedicated device driver with a standardized
interface. That way each driver would have a uniform interface
available for other software to use. 

I think the latter option is worth considering. For me, software
looks like a very good tool for abstraction and hiding nitty
gritty details. Implementing drivers for all the operating systems
and hardware we want to use is a bit of work, but then again, we
have a good bunch of people who volunteered as system programming
engineers! =)

One option for hardware would be to get some kind of a
microprocessor and just run NetBSD (or another suitable OS) on it. 
In addition to being an interesting idea in itself, it would let
us get rid of the PC part...  I was told we could take a look at
so called "suite chips", which are available as MIPS, ARM,
PowerPC, SH3/4 and Xscale architectures, for example, and could be
used for running NetBSD. 

The catch is...  how hard is it to build your own computer? =)
Probably we'll have to document at least a basic lightget design
(no programmable hardware) and an advanced lightget design
(interrupt controller, cpu or some other programmable hardware),
because the skill level in electronics each of us has is quite
variable. 

We could also consider some stuff like that sold at
www.coolkits.nl - for example the I10 - 16E would allow us to
control quite a bit of stuff over ethernet, only thing is it's a
bit expensive ($200) =)

Btw. Should we adopt some kind of a real-time communication method
for meetings, IRC for example?

Varis