[Community-news] OLPC News (2008-10-20)
jg at laptop.org
Mon Oct 20 15:06:15 EDT 2008
A weekly update of One Laptop per Child October 20, 2008
Elana Langer visited several pilot schools as well as ones that received
XOs during the ceremonies in June. In School 51, Elana found many of the
teachers were using the computers in interesting ways. A two-day
workshop is scheduled for this weekend to help advance the skills of
teachers and students primarily in Schools 4 and 20. Elana also worked
with teachers in School 4 to re-flash their computers into Mongolian.
Bulgan, a member of the core team, finished the translation of Turtle
Art and Scratch this week. The project unit is distributing versions of
the program to all laptop schools.
A Peace Corps volunteer in Hatgal is doing some wonderful work
supporting the XO in the local community. There are 200 computers at his
school and he is trying to work with teachers and students at each grade
level to strengthen their skills and bolster their self-confidence on
the computer. He has been asked to help at a neighboring village, and
will do so next weekend.
Elana updated the national donors’ meeting on OLPC’s progress in
Mongolia. This was an opportunity to thank Minister Otgonbayar in the
presence of the members of the donor committee for his swift attention
to the challenges the laptop initiative has been facing. It was also a
chance to reintroduce the need for local partnerships to the donor
community, which included ADB, World Bank, World Vision, UNICEF, UNESCO,
GTZ, JICA, Save the Children and several other organizations. The
reception to the project goals was extremely positive. After the
meeting, representatives from many of the organizations, including ADB,
World Vision, UNICEF and UNDP, approached Elana about collaborations.
Cambridge: After attending the recent OLPC learning workshop, two
Cambridge city officials discussed bringing the XO laptop to local
schools and community centers. The meeting was attended by Ken Reeves, a
city council member and former mayor of Cambridge; Mary Wong, the
executive director of the Cambridge Kids' Council; research assistant
John Clifford; and Stephanie Guirard, a resident of Newtowne Court. By
the end of the meeting, the group was "unequivocally" behind an XO
deployment, they said, and decided it would be best to implement a
laptop project throughout the city. This is an exciting possibility, as
it would provide a local deployment to show visitors, as well as a
convenient site for testing the technology under real-life conditions.
Their plans were announced Friday night at the Cambridge Kids' Council
meeting. Board members were very supportive. One parent on the board
called the idea "brilliant." The next step is for Cambridge to approve
Work also continues on developing support materials for more powerful
learning activities, such as storytelling that incorporates programming
and a variety of media.
1. For large country deployments, we have begun a more formal process of
regular meetings with the technical contacts, together with a process
for them to escalate their bugs and feature requests through an
email-based ticketing system. We expect this will help both deployment
teams who want their issues tracked with follow up, and OLPC staff who
want regular feedback from our deployments. The support group will
provide a regular summary of the results of country calls to OLPC. They
will also provide a presentation of all the information we have been
collecting for the technology roadmap planning in mid-November.
2. Reuben Caron continued working and testing with Ejabberd on
jabber.laptop.org. He worked with deployments in Birmingham and Mongolia
this week. Reuben also worked with the QA team to see how we can help
automate some of their procedures. Discussions are on-going between Kim
Quirk and representatives of Cambodia about their keyboards, as well as
with Perú on support for their upcoming conference and their list of
feature and bug requests.
3. OLPC’s Volunteer Infrastructure Group (VIG) secretary Stephan
Unterhauser has modified and extended the Debian program "meetbot" to do
automated logging of the online OLPC VIG meetings, as well as formatting
them, and posting them in real time to the wiki.
Action items, topics and agreements can be automatically called out,
indexed and highlighted.
4. The QA team (Joe Feinstein, Frances Hopkins, Mel Chua, Reuben Caron
and Kim Quirk) continued performance/capacity testing of the setup with
many XOs connected to a school server. Since the last "Weekend report",
we conducted the first real-life "Chat on school server" tests at 1cc.
We had 62 laptops connected to (and registered with) a school server.
Every laptop was running the Chat activity. 20 laptops were given to 20
people at 1cc; they were asked to actively chat with each other using
the Chat activity. We had this going for ~ 15 minutes.
After several minutes some of the laptops experienced noticeable delays
while our "children" were doing chatting (ticket ##8806). Moving a
couple of laptops ~ 60 feet from the school server (to the finance
office near the OLPC's entrance) caused these two laptops to stop
chatting even while they were still connected to the school server.
Moving them back closer to the school server restored chatting (ticket
Some of the laptops stopped collaborating completely, while being
located close enough to the school server and still connected to it
We continue testing with the school server while limiting to 50 - 55 the
number of laptops connected to a single access point. We also plan to
test other performance-enhancing configurations (including more than one
access point connected to the same school server). We also plan to
conduct performance testing in the "access point, no school server"
Mel Chua and Frances Hopkins organized and conducted a pilot usability
testing at the MIT Museum last Saturday. The feedback from the event
will allow us to help plan future usability testing (not necessarily in
the MIT Museum).
Joe and Reuben started working on the "system level" test automation
approaches, while Mel has been investigating the "software level" test
automation possibilities. Mel continued working with volunteers willing
to participate in our test efforts. She also visited Google to learn
their test practices.
5. Mel also focused on recruiting volunteers to be part of the community
test team. Newcomers, young people, and non-technical people are
especially invited. There will be both formal and informal opportunities
to learn about QA. For more information, join the testing mailing
6. The entire software development group began future feature planning
with preparations for next month’s technical mini-conference at OLPC.
The call for proposals has gone out. See the planning page for this
event and submit proposals athttp://wiki.laptop.org/go/XOcamp_2X
XO OS Software:
7. C. Scott Ananian spend the week working on the Journal. He gave a
talk at 1cc on Wednesday and prepared screencasts of his work: all the
media is available at http://wiki.laptop.org/go/Journal_reloaded (and
source code, too, for the brave).
C. Scott also helped get the ball rolling on the joint OLPC/Sugarlabs
9.1/0.84 planning meeting to be held the week of Nov 17, now called
"XOcamp2" apparently, and made a number of talk proposals for the
meeting on devel at . He hopes that lots more proposals from otherpeople
Finally, C. Scott and Sam Klein are leaving on Saturday morning for
Peru, to spend over a week there. C. Scott, SJ, and Walter Bender will
be presenting at the Open Source Jam in Lima, working with Hernan
Pachas, and trying to encourage a local community of XO developers.
8. Erik Garrison spent the week testing various hierarchical file
managers which could potentially be used in Sugar and working on UI
performance issues. To close the week he published a set of potential
modifications to the OLPC software distribution which dramatically
improve user interface performance
9. Eben Eliason participated in several meetings focused on approaches
and planning for a revised Journal. Work in this area seems very
positive so far; Eben will begin working on revised screenshots and use
case scenarios next week so design and implementation can be brought
together early in the next release cycle. He also focused part of his
efforts on the website redesign, creating a new interactive laptop "360"
and an interactive slideshow fed by OLPC's Flickr stream. These will be
featured on the main "laptop" and "children" pages of laptop.org,
respectively, to increase the visual appeal and impact via dynamic
10. Paul Fox continued the EC software merge effort after successfully
setting up a work environment with the new firmware compiler.
11. Chris Ball worked on a power management feature¹ -- when entering
idle suspend, we should set a wakeup alarm for five minutes later, and
should move to sleep mode (saving power from DCON and others) if we're
woken by the alarm. The first step was a kernel patch² to tell us
whether we were woken by a clock alarm, and the next step is to set the
alarm and act on being woken by it.
Chris also released a prototype of a Screencast activity³. This activity
allows a movie to be created using the content of the display along with
narration over the microphone; it could be useful for creating shareable
tutorials and walk throughs both for learning how to use the XO and for
learning in general. Chris is trying something new with this activity:
after getting it barely functional, he doesn't have time to polish it
and so is looking for someone to volunteer to take over ownership of it,
with no activity development experience necessary -- let him know if
12. Michael Stone, together with SJ Klein, wrote the 0th issue of "The
OLPC Journal" at http://wiki.laptop.org/go/OLPC:Journal in order to have
a good place to publish the devel@ mailing list.
XS School Server Software:
13. Martin Langhoff finished off the initial Moodle implementation for
the school server, made networking changes to accommodate large schools
using APs and battled some last issues with the build toolchain.
Douglas load-tested ejabberd, completed the OTP work (for root password
management) and the XS trust model for administration scripts.
Sugar / Activity Software:
14. Morgan Collett debugged connections to jabber.laptop.org, and tried
to make presence service more reliable in the face of network delays
seen in this setup. He worked on API documentation for activity authors,
and discussed 9.1.0 goals for collaboration.
15. Marco Pesenti Gritti wrote a proposal about API stability policy for
Glucose and discussed it in the Sugar irc meeting, he will add it to the
wiki with the changes that was agreed. He wrote a list of work items to
make Sugar window management more standard compliant and better host
normal desktop applications, Sayamindu offered to help with some of
those. He discussed the next generation Journal design and he is excited
about the perspectives, hopefully next week we will also make progress
on the UI side, with mockups by the design team. Marco fixed the various
issues when running multiple Browse instances; file pickers and
downloads are associated the appropriate window. He started refactoring
the zoom level window management logic, which is complicated and
currently messy, basing on patch by Benjamin to fix the home view
16. Tomeu Vizoso worked on the following smallish tasks:
* Added to Browse the possibility of downloading links and
images to the journal, these new options have been added
to the palettes that appear on right-click.
* Added a removable storage device icon to Sugar's frame
and a way for Shell components to react to new devices,
this is in preparation of further improvements to the
handling of usb sticks.
* Made the shell start 70% faster by not checking the
well-formedness of each installed bundle each time we
boot. These results are on a regular laptop with a
conventional hard disk, I expect the improvement to be
smaller on the XO because small non-sequential reads
from NAND are faster, but is still a good deal of work
that we stop doing at startup.
* Added previews to the text snippets placed in the
clipboard side of the frame.
* Did some inconclusive performance tests on ubifs, will
continue next week.
17. Simon Schampijer has been landing the use of gconf for the profile
in sugar-jhbuild. The profile is now using gconf to store the
preferences. The old API in sugar/profile has been kept around to not
break activities using it, for example to request the nickname or the
color of the user. You can keep on running multiple instances of the
emulator by using the 'SUGAR_PROFILE=username sugar-emulator' command.
This keeps on working since we use gconf-dbus in sugar-jhbuild and
therefore run one gconf daemon per instance.
18. Sayamindu Dasgupta worked on revising the Khmer keyboard layout so
that it adheres to the national NiDA standard as closely as possible. He
also worked on adding fallback language support for translations (eg: an
Aymara user would like to see Spanish translations as fallback if Aymara
ones are not available instead of the default English). In the Sugar
department, Sayamindu continued his work on Read and added support for
handling external hyperlinks in the underlying evince python bindings.
Laptop Power Measurements:
19. Richard Smith spent the week looking further into the power
logfiles, plus the ones coming in from the community. To date he has 53
files that have held up to his initial screens plus additional
processing. Most of the data came from builds 766 and 767, but a few
Joyrides and earlier builds were in there as well. More logs arrive
about every four hours. Some stats so far:
Ave Power range: 5.7 - 6.7 Watts
Run time range: 2.7 - 3.44 hours
Min-Max idle power: 4 - 7.7 Watts
Min-Max Wh: 17.7 - 19.5 Watt Hours
Min idle power is an educated guess. The transition from full to
discharging is not synchronous with the start of the 15 seconds
measurement period, resulting in a power reading that is not accurate.
So the min reported can be too low.
Richard is not yet able to establish a baseline. More investigation on
where that one watt of variance is going is needed plus looking at the
two Wh of battery life difference. Further tests with more controlled
conditions are probably needed. Richard is specifically thinking about a
no-wireless test where the WLAN is put into reset.
Thanks to everyone for taking the time to run the tests and to submit
logfiles - especially the people who ran multiple tests. Having data
from the same build, same machine, same battery on multiple runs is very
useful as he attempts establish a idle power baseline measurement.
Please continue. More data is good. In the process several people
learned that they had the start of the charge balance problem and were
able to take steps to correct it.
20. John Watlington discovered that the first designs for the XO
(pre-A1) included an IDE/NAND bridge (Phison 3002), which was dropped in
favor of CaFE/JFFS2. This decision is being revisited for future
hardware. The Asus EEE uses an IDE/NAND bridge, and a disk model that
John has been considering. One partition uses SLC (single bit per
transistor) for reliability and stores the OS, while a second partition
uses MLC (multiple bits per transistor) to provide less reliable user
21. UBIFS testing is on hold while Deepak Saxena and the UBIFS
developers work on identifying and fixing a bug which cropped up
immediately in John's tests. JFFS2 testing has been slow due to the
hassle of having to reboot the test laptops on an almost daily basis. He
also worked with Erik Garrison to determine some next testing steps for
22. Pre-production samples of keyboards with the thicker rubber membrane
have been installed in a small number of machines, and are being tested.
Barring any problems, these should start appearing in production laptops
23. Pre-build laptops with the new touchpad and keyboard controller have
been shipped from Quanta, and should arrive on Monday. These are now
expected to enter production in December.
24. Mitch Bradley added mesh support to open firmware for the multicast
updater. He did some performance work on the USB stack, and tried to
answer a bunch of questions from Quanta.
25. Deepak continued to work on flash file system replacement options
for the 9.1 release.
26. Guillaume Desmottes implemented the last bits of the new search
protocol in Gadget. He released Gadget 0.0.2 which should contain all
the requested features.
On the Gabble front he finished to implement the new protocol as well
and merge the new Gadget API branch. In order to drastically simplify
Gadget integration in Sugar, he investigated a new path where buddies in
views where advertised as online by Gabble. He implemented it as a proof
of concept and was able to very easily request views and making their
activities and buddies appear in the mesh view without (almost) any PS
change! He also released telepathy-python 0.15.2 which contains new API
which are needed to perform Gadget searches.
27. Ricardo Carrano has spent the week in:• tests with the XO acting as
an access point, working with students at UFF to build a wireless sparse
mesh test bedhttp://wiki.laptop.org/go/Wireless_Sparse_Testbedand
working with Cozybit on the remaining WPA timing issues.
28. Javier Cardona worked on driver support for the "wakeup on
lan" (WOL) functionality that currently is implemented in the wireless
firmware. We can now wake up the XO based on the presence of a number of
predefined 4-byte patterns in the received wireless frames, making
possible scenarios such as waking up on ARP requests for its IP address.
And in other news…
Maureen Orth has posted a touching entry about the XO deployment at
Escuela Marina Orth to the Vanity Flair blog. Maureen helped build the
school in the 1960s as a Peace Corps volunteer in Medellín.
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 140762 bytes
Desc: not available
Url : http://lists.laptop.org/pipermail/community-news/attachments/20081020/f89f98e9/attachment-0001.pdf
More information about the Community-news