[openbeos] Re: Some useful thinkings and ideas...
- From: "Michael Phipps" <mphipps1@xxxxxxxxxxxxxxxx>
- To: openbeos@xxxxxxxxxxxxx
- Date: Thu, 11 Oct 2001 20:39:50 -0400
:-) I agree that it would be bad to lose focus, but there are some
high points to this that I think are of note:
1) Memory is *DIRT* cheap. I saw 512 meg of PC2100 DDR ram for $80.
I know that you wouldn't put *that* much in an embedded box, but, say,
128 meg isn't unreasonable. At that point, what is the difference between
embedded and not?
2) BeOS is PHENOMINALLY modular. That is the premise behind the development
methodology we have been working at. I have been working on a new
ScreenSaver, for example. I can replace the existing ScreenSaver at any time
with my own. I could also turn of the screen saver all together. While Be did
not
make this super configurable (and there is no blame here - this is not trivial
stuff), it would be fairly trivial to set a box up to only use a few servers.
3) The kernel that we are using is very nice. Small, compact, efficient and
sweet.
Also pretty portable. Really. I don't think that anyone could improve that part
of the
design significantly. Everything architecture specific is split out into its
own functions.
And there aren't that many.
Having said that, embedded is out of the realm of what we are focusing on.
While I don't think that it would be very hard, we have enough to do. :-)
Be tried to enter the embedded market too soon (some would say ever
entering it is too soon). It takes a lot of engineering to fit a whole OS into
32meg
with no hard drive. But it takes almost no engineering to fit BeOS into 128
meg. :-)
>Man, we already have to write a complete desktop OS. Now we need to
>write an embedded OS too? Damn, I'm suddenly very tired!
>
>Seriously, these are cool ideas. You realize, of course, that most cool
>ideas are dangerous to project schedules ("can you say 'feature creep'?
>sure, I knew you could")
>
>Not to be a fuddy duddy, but how 'bout we concentrate on the mammoth
>task already before us and worry about conquering the world a little
>later?
>
>;)
>
>> I have some suggestions:
>> 1. Couldn't be OpenBeOS developed for the embedded market
>also? I know that it's something hard to do without the specific
>hardware to test it on, but right now I have a project to work for, and
>I need a embedded version of BeOS (I know that it exist and it's called
>BeIA, but officially BeIA is dead). So what I'm proposing here it's to
>have two versions of OpenBeOS:
>> A. Desktop version
>> B. Embedded version
>> The embedded version will be a stripped-down version of
>OBOS, with support for specific hardware. But in order to be possible
>to do a Emb version, OBOS must be made with this in mind. I think that
>a very good modularity would help a lot. A very customizable
>installation for the Desktop OBOS would be very much appreciated.
>> 2. How about the multiplatform support? If OBOS is designed
>from the beginning with this in mind, could be easily ported to a
>platform. A mean, look at NetBSD... It exists for 21 platforms, even
>for obsolete hardware and they say that it is a viable solution even
>for the embedded market. The code of NetBSD could be a source of
>inspiration, right? Just take a look at their goals and the philosophy
>behind the system (in case you didn't do it already ;-) )....
>www.netbsd.org
Other related posts:
