[sugar] Weekend - Technology Team, Details
Sat Apr 19 09:39:38 EDT 2008
*Network / Collaboration*
* The multicast filter. To avoid waking up a suspended XO at every
multicast frame received (they can be pretty frequent in a dense mesh) a
multicast filter was implemented in the firmware release 22p8. Now the
driver should announce the multicast addresses of interest. If it does not
we have problems like #6818. A driver patch is now being discussed by
Marvell and David Woodhouse. There has also been discussions (at the devel
list) on the limit of 8 mac addresses in the filter.
* The 6589 blocker. We are still fighting in the #6589 front, the fix
included in the firmware 22.p9 (that also needs kernel support) was the
introduction of a "firmware ready" event (or a 200ms wait time for backward
compatibility). This fixed the bug but caused a race condition as a side
effect, so a new firmware release - 22p10 was released (today). Relase 22p10
is the current firmware version under test now, and the next candidate to go
into joyride (22p8 and 22p9 will be skipped) and succeed 22p6.
I updated the Update.1 release notes at
for most network issues. I also included most bugs we discovered with Wad in
Peabody 100 laptop test.
* Investigated PS bugs that have been filed due to latest test beds.
* PS bugs triaging
* Wrote some improvements in PS and Telepathy to help debugging.
* Discuss 1-1 tubes protocol in Salut
* Reviewed a D-Bus tube fixed
I have spend lot of time this week digging into presence bugs that have
been filled by Wad according his observations during latest test beds
(#6880 and next).
I wrote a not-reviewed-yet Gabble fix for one of them (#6883) and
improved PS debugging output and telepathy-glib to inform us when CM are
disconnected from the bus (#6905). That should, hopefully, give us more
information on some strange and hard to reproduce bugs.
* Started my new role for OLPC, working on activity collaboration
* Started http://wiki.laptop.org/go/Collaboration_Central as a page to
track collaborative activities, developer resources and news about
* Tested Jani Monoses' packages for Sugar in Ubuntu 8.04 (hardy) which
is due to release next week as this will be a very useful way for
people to run Sugar to write activities
* The theme for this week's sugar developer meeting on IRC was
collaboration. We had a good discussion about API issues in
* Started reviewing the state of collaboration code in core
activities: submitted patches for Write for #5062 and #6021.
This week, I published two commentaries on planning and I spent many
hours working to broaden and deepen communication channels throughout
the project. I also helped prepare a collection of activation leases for
Mexico and I began trying to comprehend what packages could be updated
in a near-term bug-fix release if the need arose to make one.
Wrote up the first level of detail feedback from the sales team as to the
priority of bugs fixes and features needed. I presented a summary of what
works and diagrams of basic network configurations that have been tested and
documented. There was much discussion about activation leases and
mesh/collaboration. The sales team is also interested in easier XO upgrades,
more video/text documentation of how things work, and better battery life.
This information is part of the goals and priority planning that has been
discussed heavily on various mailing lists.
* started branching things for dist-olpc3 based on F-9 for testing
* spoke with Fedora Brazilian Ambassadors on how they can help OLPC
*General Software issues*
I have been figuring out possible ways to make the language packs more
effective. Currently, the existing language packs are only a hackish
solution to quickly check translations (and contains the translations
only for the bleeding edge development branch). For actual deployment
scenarios, we need more fine grained control over the contents of the
language packs. Some of the "parameters" which should go into the
building of the language packs (so that deployers can create a custom
language pack for their deployments) include:
a) Branch being followed (ship-2, update-1)
b) Activities being included in the deployments
I'm working on a web based tool which can do this - hopefully I'll be
able to come up with something by next week.
I also made a short presentation in the first meeting (over telephone)
of OLPC India (which happened this Friday), outlining our translation
workflow, so that people interested in contributing to the
translation efforts can join us.
- worked on getting more stuff upstream (I gave a summary of that at the
software meeting, I'll give another update for weekend once the patches are
actually in Linus's tree).
- cleaned up kernel build trees on filer (which made me discover some long
lost patches). This is still in progress, it's slow going.
* Polished visual designs for the GUI control panel, and exported
icons for its modules
* Took an extremely thorough first pass at refining the visual and
interaction design of the Journal
* Undertook a rewrite of the activity launch feedback code, providing
a new model that should improve stability in the field and emphasize
the zoom metaphor that the UI is built on.
Adam Holt, Emily Smith, and Sandy Culver continue to work hour by hour
through all the issues associated with non-deliverable laptops, returns,
refund requests, tech support questions, and just listening to people who
really need to vent.
Michael Stone and Richard Smith helped to support Mexico's deployment by
debugging and working around problems with the OLPC manufacturing data
server. This pointed out the need for better process and backups for the
activation and lease management issues.
Kim Quirk met with Bruce Lai and Tom Kambouras from NYC schools to go
through their list of technical issues. They are hoping to start the rollout
of 1500 laptops in the next few weeks. One of the requirements for all US
schools is to be able to limit the wireless access of a laptop to only those
belonging to the school.
John Watlington will be visiting Peru again next week to help keep that
deployment moving forward.
* We are moving ahead with purchase of three new servers. These are to
replace pedal, crank and xs-dev.
* Still backing up the filer on grinch so that we can assure the data that
is there before adding disks to the Coraid 1521 disk array.
* On Thursday 17th of April 2008 1cc had a scheduled downtime of
unexpectedly long duration for the upgrade of the wireless link from 1cc to
MIT from 100MB to 1 GB.
*Conferences / Other*:
* Richard Smith did a presentation on "One Laptop Per Child and Alternative
Power Sources" in conjunction with the Massachusetts Power Shift (MAPS)
event at Boston University on 2008-04-12. He was assisted by SJ Klein and
Henry Edward Hardy.
* SJ Klein and Henry Edward Hardy presented an XO to the MIT Museum for
their permanent collection on 2008-04-16. John Durant, the Director of the
MIT Museum and Deborah Douglas, the Curator of Science and Technology wish
to express their sincere thanks for our donation of an XO for their
permanent collection today. They are keen to have a few more current models
for hands-on demonstrations, and would very much like to do an historical
exhibit showing the key stages of conception, design, production and
Picture of Scott giving his talk at FISL (provided by Dennis Gilmore):
Wad and Kim visited 2 possible sites for the next home of the 100 laptop
test bed. One is the old Rounder Records building near Davis Square and the
other is a garage behind the cinema at Fresh Pond Mall near the Alewife T.
* SJ has offered to pick up the management of the Developer's Program, which
is ready to start sending out laptops. It has been a number of months since
this program was running. Please send people to http://projectdb.olpc.at/
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Sugar-devel