If your apps are Java-based, there might be a gain by using the Dalvik
VM instead of Sun's implementation. Bionic (the C library in the Android
stack) might be faster then the GNU one or uclibc (I recall it being
different, at least :P). But other then that, I'm not sure how much of
the project itself would be useful. You could look through the kernel
options specified in the stock "Android kernel", to get a feel for what
needs to be enabled/disabled too.
Jerome
On 16/03/2010 8:29 PM, Jason LaChapelle wrote:
Thanks for the quick response Jerome.
I was "afraid" the response was going to be that I'm better off using
a linux kernel modified to suit my needs. The reason I was looking at
Android was a lot of the functionality we'd need is already built into
the platform and it's been proven to work (and be speedy) on low-end
hardware (compared to desktop computers). This includes things like
Wireless, Bluetooth, and memory management. I know I can compile a
linux kernel with all these things but the draw, as it were, to
Android is (once again) the fact that's been proven to work really
well on limited hardware.
Nonethless, it was helpful input and it's probably the direction I'm
going to take. And, to be fair, it was probably the direction I was
going to take all along but I have to feel out the Android platform
first - as it always comes stock with a lot of the stuff we need :)
On Tue, Mar 16, 2010 at 7:24 PM, Jerome Duval <[email protected]
<mailto:[email protected]>> wrote:
It seems to me like you would be better served by compiling the
Linux kernel to suit your needs. The entire Android platform would
be the first thing I would strip in your project, if I'm
understanding what you want to do correctly.
Build the kernel with the drivers you need specifically and
compile whatever other libraries you need for your project (a
graphics lib if your applications require one, for example).
Jerome
On 16/03/2010 6:18 PM, Jason LaChapelle wrote:
Hi,
I've been working on a prototype for a project and I'm at the
point
where I need to start finalizing the hardware. I'm currently using
Windows Mobile for this project but I want to move to Android as I
feel it may be a better fit, provided I can achieve what I need to
achieve.
I'm building a device that might not technically be "mobile".
It won't
have a phone, or GPS capabilities, or even many of the
features found
on a phone. It will, however, run on very limited resources
and will
need the ability to access the internet and run one or two custom
built applications. Android seems like the perfect fit as it
is able
to do these things really well (internet and run applications).
My questions are - will I be able to strip out the parts of the
platform I do not need without affecting the core
functionality. For
example, there will be no phone so I would not want the phone
functionality. The ultimate goal is to make the OS as light as
possible, so removing the uneeded "stuff" is a "must-have".
Also - would the kernel be modifiable to the point to get it to
communicate with an installed USB port? I assume the answer
here is
yes, but you never know!
Many thanks in advance!
--
unsubscribe: [email protected]
<mailto:android-porting%[email protected]>
website: http://groups.google.com/group/android-porting
--
unsubscribe: [email protected]
website: http://groups.google.com/group/android-porting
