Re: Lua OS!

[Javier Guerra <javier@...>]

On Tue, Dec 22, 2009 at 11:25 AM, steve donovan
<steve.j.donovan@gmail.com> wrote:
> Javier mentioning eCos naturally suggests that this would make an
> excellent embedded platform.

right; that's the main advantage eCos brings over Linux.

On Tue, Dec 22, 2009 at 11:07 AM, Timothy Baldridge
<tbaldridge@gmail.com> wrote:
> The killer of almost all hobbyist OSes is device support. This is why
> SkyOS recently switched over to using the Linux kernel. So the
> question becomes, do we want a more flexible, simpler kernel with less
> driver support, or a more complex kernel with better driver support,
> but less "hackability"?

about device support, definitely for development/hobbyist use, it
should run on desktop machines. either on a virtual machine, or
porting the whole environment to the Linux kernel.

that's what EL/IX is, a POSIX-based common specification for embedded
systems, which (AFAIUI) happens to be a subset of Linux functionality.
 IOW, the Lua runtime would be written to EL/IX specs, so it could be
either embedded with an eCos kernel, or installed (as a process?) on a
standard Linux machine.


> Perhaps the question comes down to what the goal of the LuaOS is? So
> far the goal sounds like "this would be awesome! let's do it!"...but
> perhaps a more defined goal would allow us to determine what kernel to
> use. If we generate a list of deliverables of what the OS should
> due...then perhaps the rest will follow.

i seriously doubt that such a thing would be developed on a mailing
list.  rather, what i hope could happen is that these brainstorms
inspire someone with enough time on his hands to start developing it.
At this point, it would be a personal project, so it would have
personal goals.  Maybe this someone wants to embed into a flashlight,
or wants a big lighttraffic controller, or run in his daughter's
Beowulf cluster.  That personal goals would guide his choices,
hopefully using the best ideas thrown here.

Later, right after he shows his proof of concept code, someone else
will pop up with different goals: either to make it embeddable on
silverware, or to orchestrate the Galileo satellite constellation.
then, we would have either two different implementations of similar
ideas (like we have so many multitasking Lua projects); or maybe we
would witness an amazing stretch of scalability concepts.

wow, i'm optimist today... must be the lack of sleep

-- 
Javier