[OLPC-Philippines] OLPC Ph Jan.31 meeting

[Bernie Innocenti]

(sorry for the long delay, I have been traveling a lot lately)


Carlos Nazareno wrote:
> I'm going to be blunt here (sorry, Bernie, Mel :-/), but as we
> discussed in the developer list last year, Sugar has major problems
> interfacing with other systems because the concept of files (which is
> central to almost any existing computer system) has been removed and
> replaced with the journal system (which has its own problems, key of
> which is being flooded with unnecessary entries over time).

You should test the upcoming 0.84 release, which has a lot of
improvements in the Journal backend and UI.  The implementation is
now much closer to what was originally designed (but important
features are still missing).

You can easily preview Sugar 0.83 in SoaS.  A little less easily, you
could install beta releases of the upcoming round of Linux distros
such as Fedora 11, Ubuntu Jaunty, Mandriva 2009.1 and so on.


> Also, I am for the exploration of alternative OSes/UIs to XO OS
> (Fedora + Sugar). Although Sugar presents a clean UI, the speed at
> which applications launch is... um. slow.

Startup speed has also improved *a lot* lately.  You can improve it
even further by disabling the rainbow security daemon, which is a
feature that does not even exist in other OSes.

But the real "fix" for startup speed is switching away from the slow
NAND of the XO.  You will find that things start really quickly with
on regulr netbooks, and decently fast on an XO fitted with an SD card.


> Anyway, given all the current upheavals at OLPC, the fact that the
> XO-2 is probably now under development, and the speed at which
> hardware/software tech evolves, I'm looking more towards the
> least-common cross-platform denominators: web browsers, Java, the
> Flash Platform (which includes Flash Lite and AIR) for content
> development.

Frankly speaking, I wouldn't count on the XO-2 being available anytime
soon, or even at all.  It's not even clear who is working on the
hardware and the software.


> IMHO, it's a good route to take and will benefit more communities and
> systems as any content we create with these platforms in mind should
> be forward compatible and would run on other systems well since any of
> the apps we create will be made with a low CPU & Memory footprint in
> mind. Thus, any stuff developed with this in mind should run on Linux,
> Mac & Windows - practically everyone who has a computer! (and
> hopefully smartphones too)

Indeed.  This is why we've been working hard over the last 6 months to
make Sugar run smoothly on all Linux distributions and all available
hardware.  With some effort, it could be ported to OSX too.  However,
Nicholas idea of porting Sugar to Windows would require a considerable
amount of effort, and it's not even clear what version of Windows such
a port should target (XP is obsolete, Vista is doomed, Windows 7 is
not yet out).


> Okay, the 3 Flash-dev XO units entrusted to me by 1CC for Flash dev
> are named after the Ninja Turtles (for obvious reasons). I've got
> Leonardo & Donatello for testing Adobe Flash and Gnash with (when I
> get free time, I'll test to see if Adobe AIR for Linux will run on
> 'em). Michaelangelo is now with John.

:-)

We are in contact with Rob Savoye, the author of Gnash.  With some
funding, he could adapt Gnash to our custom requirements.  Or we could
hire anyone else experienced in C++ development to do it.


> I've also installed Teapot's Ubuntu Linux 8.10 Intrepid Ibex
> (http://olpcnews.com/forum/index.php?topic=4053.0) on Leonardo via an
> 8GB LiveSD card and have been taking him everywhere as he fits in my
> manpurse.
> 
> No complaints so far and it pretty much serves a lot of my travel
> purposes. It even runs http://processing.org (albeit a little slowly,
> pretty much the same speed as my PC 6 yrs ago), runs MP3s w/ playlists
> & DivX movies via VLC, Opera browser runs fine and dandy, Javascript,
> Flash & Java can be turned off with a few simple dropdown menu clicks
> for surfing speed.

Have you tried installing Sugar? :)

To experience 0.83, you would have to upgrade to Jaunty first.


> Another nice thing about Teapot's LiveSD card is that it's
> nondestructive and doesn't touch the XO's internal NAND so I can
> immediately switch back to XO OS just by removing the SD card after
> shutting down the unit.

Indeed.  I wish future netbooks came with no internal flash at all,
so users could download or purchase operating systems and appliances
separately.

The Gdium Liberty, an interesting product we have been shown at
FOSDEM, works like this:

  http://www.gdium.com/en/product

The company that produces them seems eager to find partners in the
education sphere.  Would you like me to put you in contact with them?


> Anyway, as I said, XO OS (Sugar) is not very suitable for older users
> (nor quite compatible with other systems because of the removal of the
> file & file manager paradigm). As such, maybe we can try and customize
> alternate Linux builds for the XO-1?

Agreed, at this time Sugar does not meet the expectations of older
students.  This is not a design limitation ("low floor, no ceiling:),
but it's a practical shortcoming of the current versions and it might
take another year or two for Sugar to offer a better experience for
advanced users too.


> I was also talking to Ed Cherlin and Yama Ploskonka and Mel Chua from
> the developer lists and I suggested that maybe the XO could also be
> used to jumpstart entire communities and help scientific projects and
> not be restricted to the just the use of education of small children.
> We also discussed possible usage to aid in electoral polling.
> 
> Maybe other customized lightweight Linux flavors could be tested/used
> for projects such as geotagging and cataloguing ecological diversity
> (the Philippines is a biodiversity hotspot)? And even cataloguing
> language, culture, music, dance, oral tradition, etc as the
> Philippines is very rich in culture.

DebXO comes to mind, if you want to try it.  The desktop choices it
offers seem more suited for geeks like me rather than students, but it
might be a good base for further work.

My sense is that, the moment you drift away from Sugar, you have to
start putting together an educational suite from scratch.

Mandriva and Canonical are working on educational flavors of their
respective distros, but for young children Sugar remains unbeatable.


> I was also talking to Gabby Dizon, president of GDAP (the Game
> Developers' Association of the Philippines) about possible development
> of content on the XO by local game developers (and not just 1-3 day
> game jams! we mean quality indie games!)

Interesting project.  You should contact SJ of OLPC, he organized a
few of these already.


> Anyway, those are some thoughts.
> 
> Ed & Yama immediately advised me to fill out a contributor's program
> form to apply for units for some of these projects as some like
> working with universities & DENR (the Philippines' Department of
> Environment and Natural Resources) for the cataloguing of Philippine
> biodiversity as results from such studies would immedately benefit the
> world science.
> 
> There's so much we can do with the XO-1 to change our country and
> change the world.

Indeed.  The tremendous social impact of OLPC is the #1 reason why me
and many others have been attracted to this project.


> I suggest we start pooling ideas for programs and usage of XO-1s for
> the Philippines, and then make combined contributors' program requests
> addressed all in one go to a single address for convenience of 1
> Cambridge Center (and as per 1CC contributors' program guidelines).
> 
> So we need an address where the dev XO units will be stored and then
> lent to developers and accounted for library-style.
> 
> As for the use of XOs in classroom settings, again, I recommend
> starting pilot programs with private schools for reasons of security,
> staff support and infrastructure. In fact, I even recommend that the
> pilot programs be done first in top-tier schools because there,
> better-educated faculty can perform controlled scientific studies on
> best practices for 1-to-1 computing as it still a relatively new
> concept on such a massive scale here in the Ph.

Good points.  Also think of setting up two development teams: one of
software engineers to make software customizations, and the other to
to produce localized contents matching the national school curriculum.

I'd like to help with the former, if we find funding for me to come.
Bryan Berry of OLE Nepal can offer a lot of advice for the latter.


> One good deployment target might be Xavier School. Take a look at this link:
> 1-to-1 Computer-Based Instruction Lesson Plan Design Contest
> http://w3.xs.edu.ph/?p=5088
> 
> Also, Xavier School has just dumped Windows for Linux ;)
> Dump Windows, save millions
> http://www.chinwong.com/index.php?/site/comments/dump_windows_save_millions/
> 
> Again, I wasn't at the meeting and haven't watched the UStream yet,
> but although jumping headfirst into the poorest communities with
> laptops seem like an incredibly noble effort, IMHO it would be slight
> shades of blind leading the blind. So again, I suggest start with
> controlled environments where 1-to-1 computing best practices can be
> developed, and then knowledge transferred to further communities.
> 
> I hope that I'm not raining on everyone's parade but although there
> have been many success stories about 1-to-1 computing, there have been
> a number of horror stories in the U.S. as well like students IMing
> each other and cheating, a school left with laptops and no idea how to
> effectively use them in a classroom setting when the faculty member in
> charge of the laptop program up and left, difficulties in making the
> laptops relevant to school curricula, and even horror of horrors:
> students using the laptops to access porn on the internet.

I would make it clear to the schools that it is their duty to ensure
their internet connection is appropriately filtered for specific
content they don't want their kids to find.

Filtering directly on the laptops would be a lot less effective,
because smart kids always find ways to disable the restrictions, and
the other kids ask them to do it on their laptops too :-)


> Furthermore, the XO or any laptop or computer + networks are simply
> delivery systems.
> What is our payload? Educational content/curriculum. Without properly
> contextualized educational content, these laptops a lot of their
> potential would be wasted.

This is why I think all these small and rugged netbooks are capable of
offering an equivalent learning experience.  The key point is the
software.


> Regarding Sugar, since Sugar has been built to be very lightweight and
> CPU-friendly from the get-go, what I'd like to see is Sugar
> installable and can be run on top of other OSes like Mac and Windows,
> very similar to http://scummvm.org or educational CDs with plenty of
> educational games and activities for the user. This would be quite
> challenging, but if sugar can evolve be transformed into a
> cross-platform runtime environment, it would really change the game.

See above for what can be already done today, and what could be easily
done with some development effort.


> As it is, typical Linux programs cannot just be launched from
> sugarized icons in Sugar, they still have to be sugarized and this has
> become a bit difficult because of the XO's Rainbow security system and
> not all apps can be sugarized. I hope the sugarization can be
> modified to be more sugarization-friendly for the many, many
> existing non-sugarized apps out there.

We've been wanting to deliver better integration of native Linux
applications from day 1, but we always lacked engineering resources to
do it within OLPC.

Now that Sugar is a community-driven project, anyone could step up and
work on interoperability.  2 to 3 months of effort of an experienced
developers should solve most of the existing issues.

As a last resort, Rainbow can be disabled to push the tradeoff from
maximum system protection to maximum choice of available applications.


> After all, XOs can be
> jailbroken and kids who have access to the XOs also have root
> control of their Linux systems. I like this ethic of "you own the
> machine, you can do anything you want with it" as it encourages
> experimentation and curiousity-encouraged creative hackery.

Rainbow has never been designed as a way to restrict users from doing
unwanted things.  It restricts *applications* from doing unwanted things.

The OFW security, however, is a form of DRM and could theoretically be
used to restrict users from doing things.  We chose instead to let
users request a developer key if they wish so, because our intent is
to block thieves from using stolen laptops, not kids from learning how
their computer works.

Uruguay wanted more restrictions on the kids, so they implemented
another security system that I have never seen.  It's open source,
they can do whatever they want :-)


> Kent, I think you used to do package building for Vector Linux right?
> I wonder how fast it would be if ported to the OLPC XO... same if a
> version of Damn Small Linux or Puppy Linux with the 2.6 kernel could
> be created...
> 
> Doc Mana! If you have time from your busy schedule, maybe you can
> modify Teapot's XUbuntu XO version:
> http://olpcnews.com/forum/index.php?topic=4053.0
> 
> Another very game-changing tech would be the ability to utilize
> Internet-enabled celphones as data modems via USB like the newer Nokia
> devices. Now since celphone coverage is very ubiqiuitous in the
> Philippines, this would bridge the digital divide between haves and
> have-nots even better.
> 
> Anyway, food some for thought.

Yeah, really a nice summary.  You should summarize the result of this
discussion and publish it on planet.sugarlabs.org or OLPCNews.


> I hope I haven't stepped on anyone's toes or rained on anyone's
> parade, but in todays toughening economy, we have to be very
> pragmatic.
> 
> Nyt all!
> 
> -Naz
> 
> P.S. the XO's wifi range is unbelievable. It's a wardriver's wet
> dream: whereas normal wifi devices pick up 3 hotspots somwehere at
> Araneta Center, the XO picks up 10. In fact, when driving in a car, it
> picks up private residence and internet cafe signals on the go :P

Yes, the wifi range is great.  Be warned, however, that the mesh does
not scale very much, and some doubt there will ever be a proper fix
for this.

Plan access points in schools.


> P.P.S. I haven't been playing hanky-panky with Leonardo, promise!
> John, I had lent Donatello to DISCS for some time, but I had to get
> him back for proper research and benchmarking because wiping Leonardo
> then installing/uninstalling Flash & Gnash was already getting very
> ridiculous.
> 
> P.P.P.S. John, when you're done experimenting with Michaelangelo,
> maybe you could lend him to Dr. Manalastas for a while. (saksakan din
> muna natin ng Ubuntu para makita mo diperensiya) Ubuntu on the XO
> really changed things for me as an "average" user. Anyway, I really
> need at least 2 units to do proper comparisons and benchmarking.
> 
> P.P.P.P.S. Jerome, Rowen, the guys at the devel list said that Ubuntu
> on XO's power consumption may have cut battery time in half because
> some of the special AMD Geode power-saving instructions hadn't been
> placed in the Ubuntu XO yet.

-- 
   // Bernie Innocenti - http://www.codewiz.org/
 \X/  Sugar Labs       - http://www.sugarlabs.org/