[Community-news] OLPC News (2008-10-06)
jg at laptop.org
Mon Oct 6 14:05:27 EDT 2008
A weekly update of One Laptop per Child October 5, 2008
Rwanda officially launched of their laptop initiative last week.
President Paul Kagame presided over the ceremony in conjunction with
Education Minister Théoneste Mutsindashyaka. The event was attended by
the Rwandan prime minister, the supreme court’s chief justice, the
entire cabinet, and leaders from civil society and the NGO community.
Five hundred students and their teachers from the first laptop schools
also were on hand. President Kagame, Minister Mutsindashyaka, Nicholas
and David Cavallo spoke to the gathering. Kagame committed his office to
bringing laptops to every primary school child in the country.
A highlight for the children came when they lifted their XOs to take
pictures of the president and discovered that they could also frame
themselves into the shots, so that they would appear in the picture with
Afterward, the Rwandan core team, along with Juliano Bittencourt, Brian
Jordan and David, led a workshop. The students developed projects
depicting their own visions for What Rwanda will be like in the year
2020. For the most part, they programmed in
Scratch, using images they photographed, downloaded or drew. The adults
were amazed by the kids’ visions, and by how much they were able to
construct in a very short period of time. The president’s science and
technology advisor was bowled over by their prowess, imagination, and
strong optimism for the future, including their own roles in it.
The Rwandan core team and steering committee participated in a separate
David and Juliano made presentations, as did Richard Niyonkuru, the
Rwandan government OLPC coordinator, Guy Serge Pompilus, the Haitian
government OLPC coordinator, Tony Earls and Maya Carlson of Harvard
School of Public Health and Bruce Baitke of Green WiFi.
David and Juliano also spoke before a session of the East African
Legislative Assembly, which was meeting in Kigali. They found tremendous
enthusiasm among the parliament members. Several made strong commitments
to bring OLPC to their countries. A few of them visited the Kakugu
laptop school in Kigali the following day, which prompted further
A day later, under the headline, “EAC MPs Want OLPC Adopted in All
States,” the New Timesof Kigali reported that delegates attending an
East African Community inter-parliamentary relations seminar in the city
called for adoption of the OLPC program by all five EAC members:
Tanzania, Burundi, Kenya and Uganda, as well as Rwanda. We applaud their
Starting October 13, all G1G1v2 XOs will be manufactured with release
8.2. The image needs to pass the final Quanta tests next week. If no
issues arise, the release is final. Congratulations and thanks to
everyone who worked so hard to make this happen! Release party plans are
in the works.
The entire software development team continues to focus on the
completion of the 8.2 release effort. The final work consists of
significant testing, especially unstructured testing to identify
potential problems in unexpected areas. This testing does tend to set
off some false alarms as the team revisits features that haven’t been
The team also began discussions on the structure and management of our
future releases, especially the next minor release – 8.2.1 – that will
focus on priority bug fixes, and the next major release – 9.1 – with
important feature enhancements. As part of those discussions we’re
trying to expand our release timeline to include additional releases so
we can better coordinate the work we need to do with our twice-yearly
major release plan.con
1. Greg Smith says we will spend the next week finalizing the
documentation (e.g. http://wiki.laptop.org/go/Release_Notes/8.2.0),
updating the wiki and preparing the release announcement. He has also
kicked off release planning for 8.2.1. The current plan calls for a
small maintenance release in November (subject to change). Targets for
bug fixes and deployments are now being negotiated.
Release 9.1 planning is also picking up. The goal is to agree on target
features and major areas of work by the end of October.
2. Joe Feinstein, Frances Hopkins, Mel Chua, Reuben Caron and Kim Quirk
as well as many community developers and volunteers tested the final
candidate builds this week for Release 8.2. Tests included
laptop-specific features, bug fix verification, and system level testing
with 44 laptops connected and registered to a school server. The ECO and
final paperwork are finished for approval this build. Next week, we will
continue with documentation and larger system tests in parallel with
manufacturing final test.
3. Reuben worked with the Birmingham deployment as they prepare for
their upcoming distribution of 14,000 XOs to 37 schools. Among other
tasks, he helped them troubleshoot a production XS.4 installation.
Reuben also worked on reviving and updating Jabber.laptop.org which he
hopes will be operational soon.
4. Adam Holt and Eben Eliason finalized changes for the stuffer sheets
that have been updated for this year's G1G1 program. Adam and Mel are
putting the finishing touches on how to upgrade to the latest 8.2 for
the Release Notes.
5. Aaron Royer, Seth Woodworth, SJ Klein, and Kim Quirk are exploring
ways to involve the community in creative banners, ads, blogs, and
informal press releases to help get the G1G1 message out this year: Give
a laptop. Get a laptop. Change the world. Seth is pulling down images
and video from our creative partners (eleven) and media from our
deployments as well as our community (photographer Mike Lee and others).
Soon most of this material will be made available to the community under
open license (CC) to remix their own posters, flyers and other
announcements about G1G1. See http://wiki.laptop.org/go/Community_media
for more information and to see some of the work.
6. Over the last two weeks, Media Modifications updated the Sugar
Almanac (http://wiki.laptop.org/go/Sugar_Almanac) with working code
examples on the use of Gstreamer for audio and video playback, how to
track the mouse, and how to register your activity to open different
mime types. Faisal Anwar continues to document best practices for using
the presence service. He answered developers' questions such as: How do
I setup a D-Bus Tube? How is data shared between activities through a
D-Bus tube? He is preparing entries about Stream Tubes based on working
code developed and tested on XOs. Please continue to contribute to the
Almanac, a resource which shows how to put the right pieces together to
build great sugar activities!
7. Henry Hardy reports that we had 22,000 unique visitors to
wiki.laptop.org on the 28thand 29thof September. This, and a bug in a
newly installed wiki extension, caused serious performance issues on the
wiki and affected other services such as web servers and email hosted on
pedal.laptop.org. Access from the largest bloc of Uruguay addresses was
temporarily shut off in consultation with Emiliano and his team while we
worked on a solution.
Full service was restored on Thursday, October 2. C. Scott Ananian built
a squid reverse proxy server on weka.laptop.org which has led to greater
efficiency in handling wiki requests. Scot, along with Michael Stone,
Erik Garrison and Chris Ball also devoted substantial efforts to
testing, organization, diagnosis, and bug-fixing work for the 8.2
release. Some individual investigations were done into post-8.2 release
XS School Server Software:
8. Martin Langhoff continued work on installation and configuration of
Moodle on the XS school server. Douglas Bagnall continued work on
characterizing the performance and capacity of the Ejabberd component,
an important step for planning larger school deployments.
Sugar / Activity Software:
9. Sayamindu Dasgupta spent a significant part of this week isolating
and fixing a few bugs in the language pack builder. Thanks to the
members of the localization team for the testing they have been doing
with the packs. A new effort to translate Sugar into Swedish has
started. Thanks to Mattias Ohlsson for taking the initiative. Sayamindu
also worked on the Khmer keyboard layout, trying to understand its
special features, and figuring out the changes that are needed in the
various OLPC packages in order to support the layout completely. He also
added the Moon and PlayGo activities in Pootle so that they can be
translated by our translations team. During the weekend, Sayamindu
worked on an Image Viewer activity with basic zoom/rotate capabilities.
The first release of the activity is available at
also helped Kushal Das, a volunteer from Pune, India get started in
Sugar actvity development. Kushal has expressed interest in maintaining
the Jukebox activity in his blog post
10. Simon Schampijer worked on the move to gconf to store Sugar
settings. Memory consumption on the XO looks good from a first glance.
The old profile will be converted on update and the old profile API will
be kept around during the transition phase.
11. Marco Pesenti Gritti started to work on the sugar shell refactoring
for 0.84. Lots of code cleanups and some fixes. The buildbot is being
very useful for this work, as it keeps complaining about the things as
we go. Marco also met with Benjamin and Riccardo about the icon caching
strategy, we are considering several approaches, which has potential to
consistently improve Sugar graphics performance. He started reviewing
Tomeu simple datastore and had several conversations around Journal
design. In parallel he kept an eye on 8.2 release blockers, in
particular he fixed a CPU "leak" for each activity that was opened and
then closed. Finally he continued to debug Browse memory usage and found
that there is no regression compared to Update.1, which is a start.
12. Tomeu continued work on the datastore replacement, basing on good
feedback from Benjamin Schwartz and Marco Pesenti Gritti. The design has
been further simplified, contributing greatly to increased robustness.
Some discussion has started about the convenience of adding the notion
of versions to the Journal for the next release and the different ways
of doing so.
13. Eben Eliason spent Monday working with Adam Holt et al. to finalize
the design and copy for the instructional insert and letter, and
produced final drafts of the documents ready for print. He then
participated in meetings regarding the consolidation of the OLPC
websites and their re-envisioning as both highly visual and dynamic
spaces designed to attract the attention of casual browsers and
potential G1G1 participants.
Eben also continued to refine goals and designs looking forward to 9.1,
including in depth discussions of the Journal and Datastore,
particularly with regard to versioning and tagging. He hosted another
open design meeting which delved into the current ideas for the bulletin
board activity and overlay activity chat, in attempts to determine what
possibilities may exist for early implementations of these features.
12. Guillaume Desmottes tracked and fix memory leaks in Gabble's OLPC
source code. The Gadget branch was finally merged to master. He
discussed with Eben and Simon how Gadget should be integrated into
Sugar. One of the missing feature of the current API was the ability to
perform activity searches based on properties and participants at the
same time. After some discussions with others Collaborans we agreed to
refactor the Gadget/View API to use the recently merged Requests
interface (aka the Requestotron). This should make the API more coherent
with the rest of the Telepathy API and really more extensible for future
features. He also made some improvements in Gadget and the sugar
Walter Bender's Sugar Digest can be found at:
14. Erik Garrison spent the week researching post-8.2 work projects.
These include UBIFS testing, libertas thin-firmware based mesh
networking, X composite, grab key functionality, and datastore and
Fedora Classic Desktop:
15. Jeremy Katz released an updated test build of the Fedora/XO image,
based on integration with the F10 beta release. This was a good step
forward, but it also identified new and substantial issues. We’re
preparing for the launch of 100 or so volunteer testers next week to
help accelerate the testing and development work on this project.
16. Mitch Bradley got multicast NAND updating to work directly from OFW
(no need to boot a special kernel). This permits a deployment to
simultaneously update a large number of laptops with minimal effort.
There's still a fair amount of work to do before it will be ready for
prime time. Outstanding issues include performance, security, and the
17. Richard worked on support for the new keyboard controller, which was
changed at the same time as the touchpad. At first he believed that the
EC was not going to be able to communicate with the keyboard controller
while the main CPU was off. This seemed to mean that to determine which
keyboard controller was present required either Open Firmware
modifications or more invasive EC code modifications.
However, at the end of the week Richard discovered while talking to John
Watlington that some missed EC configurations were preventing the
communication while the host CPU was off. No system firmware
modifications will be necessary.
In the meantime, Mitch has released a new firmware (Q2E19) which will be
used in the laptop prototype pre-build this coming week. We hope to have
laptops with the new touchpads at 1CC in just over a week.
18. Testing of alternative solutions for increasing the storage on the
laptops continues. This week John added four more laptops testing the
Sandisk SD cards. These are so fast that they have already overtaken all
other devices under test in the amount of data written to the device in
an attempt to wear it out/generate errors. Console logs are being
obtained from a kernel crash which has been occurring on laptops running
JFFS2 (both build 8.2-760 and 656), to aid in debugging.
19. Ricardo and Ashish debugged association issues that are manifesting
in saturated spectrum conditions. They are currently testing potential
improvements and the potential trade-offs involved with operation in
normal spectrum conditions. Ricardo also worked on updating driver and
firmware documentation on the wiki, and setting up his wireless test bed
at UFF in Rio so that he engage the students there in XO projects. *
Worked on diagnosis and tested possible fixes for current wireless
issues: wpa, scanning and resets of the wireless system (#7825, #8666
and #8667).Ricardo updated some pages in the wiki regarding the wireless
subsystem and projected a testbed for sparse network that is being
installed at a University in Brazil (UFF)
Though we are convinced that the key installation time was the root
cause for some of the association problems to WPA access points (#7825),
there are still other less severe problems in associating. Currently,
we're testing fixes for a scenario where that the scanning routine is
failing (#8667) and investigating what may cause the wireless subsystem
may fail under hostile spectrum conditions (#8666).
The UFF network testbed will study the (1) feasibility of a mesh network
in connecting people in their houses and offices, (2) the effects of
mobility, (3) validate connectivity models as the gateway mechanisms,
like MPP or MAP, designed to connect XOs to a wired network and (4) will
support a lot of student projects at the University.
20. Deepak Saxena attended an Embedded Linux Developers’ Conference and
gave a keynote on "Linux Power Management Challenges for the XO and
And in Other News…
Nikos Passalis, a senior at Neapolis High School in Thessaloniki,
Greece, won the special EIROforum CERN award in the 20th European Union
Contest for Young Scientists, which was held this year in Copenhagen.
Passalis’s winning entry was a program that utilizes XOs to form
distributed systems. According to the official contest site, the
program’s “most important features are: the remote control of the
laptop’s projects, its fully automated operations without involving the
user, the great fault tolerance combined with self-fix features if a
problem is detected and the smart energy management. The modifications
to the computer’s software are very few, so it is easy to use and
functionalities are untouched. Also, this makes its application easier.
Moreover, features like virtual partitions in Ram memory, buffering
techniques, temperature monitoring, etc. were implemented in order to
reduce the computer’s hardware load to minimum and not affect its
expected lifetime. Finally, the program was optimized in order to
minimize the computer’s load during its operation. "
More at http://www.eurocontest.dk/
Brian Berry sent a link to an early evaluation of the OLPC project in
Jim Gettys <jg at laptop.org>
One Laptop Per Child
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 303900 bytes
Desc: not available
Url : http://lists.laptop.org/pipermail/community-news/attachments/20081006/e6028645/attachment-0001.pdf
More information about the Community-news