Michal Suchanek wrote:
> On 08/12/2007, Robert Dober <robert.dober / gmail.com> wrote:
>> Maybe it was an error to develop all together, the OLPC should be a
>> hardware platform, then various different Linux/*BSD based platforms
>> could be implemented ...
>>
> It is generally developed that way - you have a piece of hardware, and
> you can put different stuff on it.
> However, they needed a complete solution so they also modified a Linux
> distribution to work well on their hardware.

Since it's an open project, nearly all of the development history, 
status, design decisions, philosophy, debate, etc. is captured in 
"readily available" documentation on the Internet. The hardware platform 
is what it is for the moment, and no announcements have been made of any 
"growth path" or "upgrade path" for the hardware. Given that the 
customer base is young children in developing countries and the 
projected life span of one of these devices is five years, I can see 
where children will outgrow these, hand them down to younger children, 
and need something with a little more horsepower and industrial strength 
languages like Ruby.

The software is another story -- Microsoft has an active project to run 
Windows XP on the device, although I think when the smoke clears, that's 
going to turn out to be impractical. Among the Linux distros, Fedora was 
chosen because Red Hat and the Fedora project contributed resources to 
OLPC and Debian/Ubuntu, Novell/SuSE, Mandriva and Gentoo didn't.

I've spent enough time playing with virtual machines running the OLPC 
software that I think I can say that there would be no advantage at all 
to switching to another Linux distro. In particular, anyone who wants to 
tell me that Ubuntu is in any way superior to Fedora relative to OLPC is 
going into my kill file. (For that matter, that isn't just relative to 
OLPC). :)

Now *BSD may be another thing entirely. I don't have enough knowledge of 
*BSD to know whether it would be superior to Fedora in the context of 
OLPC. An awful lot of low-level hardware/driver work would need to be 
done, and I don't think the BSD community has any strong financial 
backers that could make that happen. But there *is*, I think, a small 
window of opportunity here for Solaris, now that it, too, is open source.

> Generally I wish there was something similar but with a decent disk
> size. Or something more like normal laptop/palmtop with the screen
> they used for OLPC.

Well, there are such things, and as far as I can tell, they're big 
sellers (and sellouts) for this holiday season. I do know of real 
working adults who have joined the Get One Give One program to obtain 
one for work use, feeling it is superior to these other alternatives.

> If you really wanted something on OLPC and it did not fit on the disk
> you could probably put it on an usb keychain. I guess it's what they
> were also thinking when making the disk so small.

There are three USB ports and a Secure Digital slot, so offboard storage 
is not going to be a problem. What I don't know is how, without going in 
and hacking partitions at the Linux filesystem level, whether one can 
make the space available for resident software larger this way. As far 
as I know these are simply data/media storage capabilities, not an 
increase in resident software space.

But I do know how to do the partition hacking, and I believe the devices 
will boot off of a USB disk, so you could in fact run anything that will 
fit in 256 MB of RAM. I'm not planning to do that, though. Right now, my 
plan is to install only three other languages, R, Steel Bank Common Lisp 
and some version of Forth. Anything else I put on the machine will need 
to be cross-developed.