OLPC News (2008-04-12) - Addendum

[Kim Quirk]

Additional community news that didn't make it into Walter's email:
*
Software Development:*
Guillaume Desmottes reports that this Salut branch
http://monkey.collabora.co.uk/telepathy-salut-cerebro2/ now abstracts all
the Avahi code. So we could have a Cerebro backend re-implementing the
needed bits and use it instead of Avahi.

A nice side effect of upgrading loudmouth to latest release was to fix a
crash when Gabble connects through a proxy (#6452).

Guillaume has fixed latest know bugs/leaks in the Salut refactoring branch.
It is now waiting review from Sjoerd. He released Salut 0.3.2, upgraded
Telepathy stack in Jhbuild and filed tickets to do the same in Joyride
(#6858, #6859, #6860, #6862, #6861, #6873). He also continued to investigate
Jingle protocol and its implementation in Gabble to prepare futur p2p stream
tubes (#4047).

Morgan Collette has been working on the Chat UI per Eben's mockups; he
deployed the latest Telepathy releases in joyride; and he has been working
on Chat interoperability with non-Sugar Jabber clients, to handle incoming
chat connections and launch Chat or display an invitation to launch it.

Simon Schampijer has updated sugar-jhbuild to use the xulrunner beta 5
sources
http://developer.mozilla.org/devnews/index.php/2008/04/02/firefox-3-beta-5-now-available-for-download/He
was hoping it would fix bugs like #6825, but it did not.

Andres Salomon worked on absorbing design docs on jffs2, ubi(fs), and logfs;
pushing more stuff upstream, meetings, and interviews.

Michael Stone worked with Scott, Chris, Marco, Tomeu, and Simon on
preparations to draft a convincing release plan describing our goals,
resources, intended use of those resources, and fallback plans for the next
development cycle.

Michael also discussed sugar architecture with Jameson Quinn, interviewed
two candidates, followed up on an internship proposal, and prepared a new
Rainbow pre-release snapshot
[http://teach.laptop.org/~mstone/rainbow.rpm<http://teach.laptop.org/%7Emstone/rainbow.rpm>]
which should be substantially more compatible with legacy software. (It
makes $HOME == $SAR/instance; thus restoring function to software which
naively expects $HOME to be
writable.)<http://teach.laptop.org/%7Emstone/rainbow.rpm>

Sayamindu Dasgupta reports that the website translation project is up and
running.  We have decided to use Git and Pootle together to manage the
website translations, to ensure easy reversal to older versions if required.
I will be also migrating the other non Git hosted translations (eg: those
for Record activity, activation server, etc) to this new system, as adding
version control makes the overall workflow more foolproof and stable. Apart
from the new Sindhi translation project, I'm glad to report that after a
long period of inactivity, the Hindi (hi) translation efforts have picked up
steam and they have almost completed the translation of XO Core (
https://dev.laptop.org/translate/hi/xo_core/). Same goes for Marathi (mr),
and XO Core for Marathi is 62% complete now (
https://dev.laptop.org/translate/mr/xo_core/).

*Testing:*
John Watlington and Giannis Galanis spent more time testing our network
performance with a number of laptops attached to a school server through
conventional WiFi.   Their results were promising, as the collaboration
reliability problems seen when laptops collaborate directly,
or connect to a school server over a mesh were minimal (specifics at
http://wiki.laptop.org/go/Collaboration_Network_Testbed).  The results vary
widely depending on the model of access point used.   A collaboration with
Nortel on the problems we are encountering with mesh networking is in the
process of beginning, with a second meeting on Monday to discuss
possibilities.

While the collaboration and networking testbed in Peabody has been
essential, its usefulness has been hampered by a dearth of public
transportation. We are currently looking for a small space in the Boston
area, accessible via the subway, which will provide a longer term home for
the testbed. Kim Quirk did preliminary wifi measurements at three possible
sites this week with good results at one in Davis Square. A more detailed
review for that site will be conducted next week and the search will
continue.

Ricardo Carrano reports that firmware release 22.p6 passed initial tests and
will be included in joyride for continued testing. It gives us better
control over probe requests with the ultimate goal of improving scalability
by freeing air time. This firmware also addresses bugs #6600, #3733 and
#6528.

Firmware release 22.p8  introduces dynamic mac contention window adaptation,
that will improve reliability in dense mesh scenarios (like a
school) by reducing probability of collision. It also addresses important
bugs like #4927 and #6709 (confirmed) and #4616 (yet to confirm). This
release is under test right now.

We are very close to fixing the blocker #6589. Initial tests indicate that
we did. Two causes were identified and fixed both in driver and in
firmware.

*Multi-Battery Charger:*
Lilian and Richard worked on improving/debugging the firmware for the
multi-battery charger. Richard is seeing some sporadic cases where the
firmware seems to hang.  Lilian is unable to reproduce the problem.

*Conferences/Other:*
International Development Night, MIT Museum, 20080404 - Henry was the OLPC
representative at this event. He set up the display with the backdrops that
OLPC had AMD make for World Economic Forum. He took his G1G1 (703 with
Mexican activities) and two Spanish keyboards. Thanks SJ for pointing out
the volunteer-made handouts from our wiki and Nia and Tracy for helping me
get the backdrops.

Response was very good. Topics ranged from general plans for the future and
philosophical issues about education and development to specific questions
about G1G1 order fulfillment, plans for Windows compatibility and questions
about the hand crank and why isn't it really $100. Several people stressed
that they felt that Windows compatibility was essential for encouraging
adoption in the Third World and China, while others referred to XP as
"bloatware" and questioned whether XP would fit into our 1 GB of system RAM.

We plan to donate a MP XO to the MIT Museum for their permanent collection.

*Sysadmin:*
Henry Hardy reports that our wireless link from 1cc to internet to be
upgraded April 17. There will be a "short" network downtime associated with
this activity. This should not effect our mail, web, wiki, and development
services on crank and pedal at the colo. Other services on machines such as
update, activation, download, xs-dev, teach, thinker, will be off the net
during this downtime. Local internet services will be interrupted.

Henry is ready to move forward with the procurement of the pedal, crank,
xs-dev replacements. He has begun backing up the data on the coraid 1521
filer. Once the system is backed up and the new disks are installed we will
have another 7 TB of space for local backups, file services etc. This will
probably fill up very fast so we should think about scheduling procurement
of another (I believe this was budgeted).

Scott Ananian reports that download.laptop.org now has a 1TB disk, and
plenty of space. Working on grabbing historic builds from
olpc.download.redhat.com, replacing archived recent builds which had to be
moved for lack of space, etc.  We now have plenty of space for activity
packs, school server builds, etc.  Let me know if you want to move stuff to
d.l.o.

*Support*:
Adam Holt, Emily Smith and Sandy Culver continue to plow through the donor
services questions. Now that all the laptops have been shipped for the G1G1
program, the next 2 months will be focused on laptops that were
non-deliverable, dead-on-arrival, or 'other' problems. It is usually these
'other' problems that take so long to investigate. We had a good discussion
with out partners to understand what processes were broken and thoughts on
how to fix them.

We continue to work on issues holding up our ability to ship out for the
developer's program and for the spares and repairs program. Hopefully the
blocking issues will be resolved early next week.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.laptop.org/pipermail/devel/attachments/20080412/57410428/attachment.html>