OLPC News 2008-03-08

[Walter Bender]

1. John Watlington, Walter Bender, and Edgar Ceballos (Brightstar)
spent this past week in Huampani (Lima) participating in Peru's first
train-the-trainers workshop. (143 of the participants will be staffing
regional support centrals scatter throughout every corner of the
country.)

Using a custom image prepared by Chris Ball and Michael Stone, the
teachers made great progress throughout the course of the week. They
are all completely comfortable with the UI, the Journal, and the
pedagogy. They readily navigated the various issues associated with an
overloaded network and as was made clear in a Q&A session towards the
end of the week, they really appreciate that everything on the laptop
is open (and free as in speech); and they appreciate the mandate to
take both ownership and the responsibility that comes with it.

They got deep into programming—for most, their first such
experience—developing projects with Turtle Art, Scratch, and Etoys.
(No Pippy hackers yet, but there was certainly interest expressed.)
The workshop culminated with a Media-Lab-style open house, where each
teacher demonstrated something they did with the laptop during a week
of Constructionist learning.

Meanwhile, the Peru support volunteers met again Wednesday night and
decided among other things that they will contact local universities
to organize an event of their own around developing networks of
educators and creators working with OLPC schools.

2. Antonio Battro was a keynote speaker at the Las Vegas International
Conference of SITE, the Society of Information Technology and Teacher
Education. Some 1,300 teachers and educators from many countries
attended the Conference. Antonio's talk "OLPC: the cognitive
challenges" had one of the largest audiences and was followed by an
engaging discussion. The interest and the will to participate at OLPC
was significant among the participants.

3. EC: The saga of the Embedded Controller continues. The code that
handles shuttling bytes from the touchpad and keyboard up to the host
has problems if the host does not read the data fast enough. This
might be at the root of some of our touchpad problems. Fixing this
will require a significant rewrite. Richard Smith feels he now
understand
things enough to do the rework, but it will not be available for Update.1.

4. Multi-battery charger: Bitwork's has assembled a new PCB in and it
is functioning as expected. Richard and Lilian Walter added some
hardware debugging routines to the firmware to assist Bitworks in
their testing. We are still awaiting the larger parts back from the
reworked tooling.

5. Q2D14 firmware release: Richard pulled a few new fixes from Mitch
Bradly into the current release tree of the system firmware and is
preparing to release a Q2D14. The primary reason is to fix an issue
with forced upgrades. Once an upgrade has been downloaded, it will be
installed on the next reboot. But if the firmware is being upgraded,
that reboot will hang unless an external source of power is available
(Ticket #6245). The suggestion is to either defer the firmware upgrade
until power is available. This will hopefully help in places such as
Peru, where many laptops will be off the power grid, when they mass
upgrade machines.

6. Battery failures: Richard finished his analysis on the six
batteries that were provided to him from laptops returned with
charging problems. He's satisfied that its not a charging problem with
the laptop circuitry and will be returning the batteries back the the
manufacturer for deeper failure analysis.

7. School server: John Watlington reports that after last weeks
testing, the current advice to trials and early deployments about
wireless interconnection in the schools is the following:

* Up to 10 (hopefully 20) laptops will work fine without any
infrastructure; Groups of children away from school can share and
collaborate;
* Up to 40 laptops are supported either by using a single access
point, or a school server with one Active Antenna;
* Between 40 and 80 laptops are best served by a school server with
two active antennas;
* Between 80 and 120 laptops can be served by a school server with
three active antennas;
* Over 120 students, the schools need to move to more traditional
access points to support the network loading.

We will be setting up a longer term (and larger) mesh testbed in a
location near Boston where the RF environment is less variable than
1CC, to continue testing improvements to the software. We hope to
improve the above numbers! (Chris Ball and Dafydd Harries tested
laptops in mesh mode against a school server running a Jabber
server—with 32 laptops connected to the school server, the laptops
were all able to share a Chat session  and a PDF file between
themselves without any failures.)

8. Peru connectivity: As mentioned above, John accompanied Walter to
Peru in order to set up servers and network access for participants in
their regionals leader training sessions. The sessions were held at
Huampani, a small resort area about 25 km from Lima up the Rimac river
valley. Four VSAT connections were set up to provide internet
connectivity to the sessions. While John arrived expecting problems
with the presence service and wireless network congestion, the real
problems had were due to strange DNS interactions with the VSAT
modem's transparent DNS proxy and the Peruvian Ministry of Education
DNS service. A fix is being tested.

9. EJabberd: We continue to have problems with obtaining a working
ejabberd build with the latest fixes. Collabora is working on this,
and as soon as it is available a new build will be announced on the
server-devel mailing list.

Meanwhile, Morgan Collett has added improved documentation for setting
up ejabberd from source for community jabber servers to the wiki
(http://wiki.laptop.org/go/Installing_ejabberd); so far one community
server (for Chicago) has been successfully set up using these
instructions, with more in progress
(http://wiki.laptop.org/go/Community_Jabber_Servers).

10. Active Antennas are now available for developers, and will be
shipped out soon to people who have already requested them. (See the
OLPC wiki for further details.)

11. Activities: Chris Ball made Pippy collaborative, such that joining
a shared Pippy activity now gets you a copy of the source-code buffer
of the host at the time you joined.

One of the newest activities developed for the XO laptop is called
StarChart. It was created by hobbyist, David Wallace. Dave received
his XO via the Give One Get One program at the end of 2007 (See
http://wiki.laptop.org/go/StarChart).

Qirat Activity Version 1.0 is complete and runs as sugar Activity.
Waqas Toor is working on making the Surahs (chapters) more presentable
when drawn on the Sugar UI canvas.

12. Presence service: Guillaume Desmottes implemented and tested flow
control in Salut stream tubes (Ticket #6647). As discussed with
Polychronis Ypodimatopoulos last week, Salut should be able to use
different backends to announce and discover services. Then we could
switch from Avahi to Cerebro. That's a lot easier than writing a new
Cerebro connection manager as we can reuse the muc and tubes codes
(and all the Telepathy interfaces). So Guillame started to design an
abstraction layer that will be implemented by Salut and Cerebro
(Ticket #6658).

Guillaime also did a few tests and reconsidered Ticket #6585 (PS must
reconnect server_plugin when NM changes IP addresses). We agreed with
Morgan and Sjoerd this was not a problem and closed the ticket.

Morgan worked on Ticket #6572 (Replace key with hash to reduce Avahi
TXT size) as it currently breaks "friending". The #6572 issue is not
as straightforward as it seemed. We derive participants' JIDs from
their keys, and friend them using the key as the identifier. So the
current patch is for testing the impact on network congestion of a
shorter value there but the solution would be more intrusive. (If the
current approach doesn't actually improve network performance, it is
not worth rearchitecting how PS identifies buddies—which is done using
their keys—so testing would be helpful.)

Dennis Gilmore and Dafydd Harries spent time this week trying to get
ejabberd running with updated patches from Process One. They had
trouble with compilation problems, patches not applying, etc.; also
with reproducing the builds that we are currently running. It turns
out that the problems were due to newer builds being done with a newer
version of Erlang. Dafydd may finally have a version that works, but
it needs testing to be sure.

13. Sugar UI: Eben Eliason made some new designs for Journal, Home,
and Frame, which he posted to the wiki. New designs for toolbars and
bulletin boards are in progress.

The most recent designs have been posted as "slide shows" on the wiki
for review and feedback. The first three available discuss plans for
future activity management, including a repurposing of the Home
circle, a completely reorganized Frame with clearer intent and support
for notifications, and a brand new Journal which introduces action-
and object- centric views, improves support for visual browsing, and
offers a friendlier interface into ones interactions with the laptops.

We are in the process of generating new designs for toolbars as well
as beginning to consider future designs for the bulletin boards (and,
more near term, designs for standard file transfer). All of these
designs are or will be posted to wiki.laptop.org/go/Designs. We
encourage those interested to review the images and descriptions
posted, and to provide feedback within the discussion pages or on the
mailing lists, as appropriate. Thanks!

Tomeu Visozo has been working implementing on the home view redesign.
The main changes have been completed and the code is being reviewed.
Next step is to discuss
the work done so far with the design team and agree on the several
details left to work through.

14. Kernel: Andres Salomon did further work in the lxfb/gxfb drivers.
He got suspend/resume working properly, prepared more patches, etc.

15. Update.1: Scott Ananian reports on the current "state of update.1"
including a discussion of the "core activity" changes at:
http://lists.laptop.org/pipermail/devel/2008-March/011537.html
http://lists.laptop.org/pipermail/devel/2008-March/011538.html
(devel archive seems to be broken? split his message in two.)

16. Localization: Scott prepared translations for the activation
server (including translations for the developer key request form)
(See https://dev.laptop.org/translate/projects/act_server/).

Sayamindu Dasgupta has started the migration process to a newer
version of Pootle and fixed a few bugs in some of the helper scripts
that he uses to manage Pootle.

Tomeu has been doing more work on keyboard bindings for the shell and
activities in light of localization.

Waqas Toor reports that the Tahoma font seems most suitable for Urdu
on the XO laptop, as Nafees font has some problems when it comes to
binding Urdu characters in smaller font size. For Pashto and Dari, we
have found Pashto Pasarlai font suitable yet, but it has some
problems. Usman Mansoor "Ansari" and Sohaib Obaidi "Ebtihaj" are
looking for other font options that contain the complete Pashto and
Dari character set on Linux platform.

Mako Hill finished a localization patch for the library, implementing
a check for localized start-pages when viewing library collections.
The next update of library-core should have at least English and
Spanish locales. (While he was in Lima, Walter finished building the
Getting Started Guide bundle for Peru. Many thanks to Edgar for the
help with the Spanish translation.)

Oz Wilder and Alon Carmeli of Babylon Dictionaries started work two
weeks ago on 16 language translations of a basic 2500-word dictionary,
to expand our current collection. They are now 99% complete, pending
final proofreading, and can be seen on the wiki
(http://wiki.laptop.org/go/Dictionary). Zdenek Broz is working to
include these in our multilingual dictionary bundle.

17. Sugar-control-panel: Simon Schampijer is working on the GUI for
the sugar control panel.

18. Support: Adam Holt reports progress on a new "Projects DB"
Developers Program with Aaron Kaplan et al. in Vienna. (Jim Gettys has
been a huge help laying this out.) Usability and features are
improving towards release, hopefully later this month.

Alan Claver helped with countless HW / pre-RMA support tickets.
Volunteers are doing so much great work behind the scenes we cannot
even keep track of it all. Adam shared Scott's deployment maps (with
disclaimer) (See http://dev.laptop.org/~cscott/stats-20080201.html).

Adam organized another Sunday Support meeting (Sunday's at 4PM EST)
with Enkhmunkh Zurgaanjin from Mongolia, who spoke about their ongoing
deployment; several new volunteers participated. Yani Galanis will be
our guest speaker this coming Sunday; topic: our wireless/mesh
testing's results/outlook!

19. Help wanted:

* A request has come in from Cambodia for help in setting up mysql in Khmer;
* We are looking for someone to "Sugarize" the GCompris icons;
* Girlstart is looking for graphic artists and designers to make their
games attractive through professional graphics (Please see the project
website http://wiki.laptop.org/go/Project_IT_Girl).

20. CeBIT/OLPC Deutschland: Holger Levsen and Christoph Derndorfer
both reported on the OLPC presense at CeBIT, the largest IT show in
Europe.

Holger reports that "OLPC Deutschland" had its first "real life"
meeting at the meeting (See
http://www.heise.de/newsticker/meldung/104668), just after having
decided by voting via mailinglist
(http://lists.laptop.org/listinfo/olpc-de) on the final name of the
project three days earlier. The group is planning to have a workshop
weekend in Berlin this spring, to develope its visions and structures.
A longer workshop which children and teachers called summer camp is
also in its early planning stages.

Christoph reports that the feedback at CeBIT has simply "blown us
away". They had hundreds of people at their booth: the main reaction
always along the lines of "I've read about the OLPC project for two
years, it's great to finally have the chance to actually look at and
use one". The most common questions "where can I get one?", "are they
already being produced and used somewhere?" and, of course, "where is
the crank?" (See http://olpcaustria.soup.io/).

21: Debian: Holger also reports that sugar 0.79 has been uploaded by
Jonas Smedegaard to Debian today, fixing Debian bug #444021. The
packaging consists of three source packages: sugar-base, sugar and
sugar-artwork.

22. Pakistan: Dr. Habib Khan reports that they had two very educative
events this week as part of our awareness campaign: The first one was
introducing OLPC to street children in Islamabad's vicinity. The aim
was to observe the reaction of children who have never been to a
school. Habib and his team spend half a day with these children and
this generated enough curiosity in the community–mostly Afghan
refugees. Though out of school, these children were quite smart and
quickly learned to operate the camera and play around with music.

The second event was an announced Open Air workshop for graduate
students of IIU. Twenty students had registered to participate in this
event and more than 50 students showed up and remained throughout.
Salman Minhas, Waqas and Sohaib gave demonstrations and assisted
students with in understanding the 20 XO laptops that we had readied
for the workshop. We expect volunteers from this lot to come forward
and help us in the localization and software projects.

23. IDCL: Ben Bederson, Tim Browne, and SJ Klein reviewed what they
need to integrate the International Digital Children's Library (IDCL)
server infrastructure (Tomcat, mysql, and some custom scripts) into
our builds.

24. GSoC: Our Google Summer of Code organization application has been
sent in. Now we need more mentors and mre project ideas. Please add
ideas to the [[Summer of Code/Ideas]] page in the wiki and contact
either Michael Stone or SJ if you are interested in being a mentor.
Also, please start spreading the word that we are looking for SoC
students this summer—students are welcome to add their own ideas to
our list.

-walter

-- 
Walter Bender
One Laptop per Child
http://laptop.org