A note from the editor:<br><br>I have been reading the discussions this past week on the mission and direction of OLPC and OLPC's outreach to the community. I'd like to reiterate that we at OLPC are committed to create Sugar as an open source project, as it provides a great opportunity for both learners and for contributors. Sugar in combination with the XO hardware extends this opportunity even further through the mesh, power management, and robustness features of the hardware required for many of the environments where we want to provide help. We have sold or given away hundreds of thousands of laptops running Sugar on GNU/Linux on the XO. We plan to continue to do that. This year's Tech Team budget allows us to almost double our personnel in development, test, and support. We are very excited about being able to hire at this level over the next few months. <br>
<br>OLPC has had an amazing amount of contributions from volunteers in the community in all areas including development, test, support, translations, documentation, marketing/messaging, and even business expertise. We truly value these contributions. What we haven't done well (at least recently) is to support people who are trying to help us. We need to fix our communications with volunteers and community groups. We need to be open and transparent without compromising the confidentiality of others that we might do business with. This is difficult but it is not rocket science. I think we can do it. Sometimes when everyone is overworked, it is much easier to focus on the details of the day than to see the bigger problems. I intend to make some specific recommendations towards better communications in the next few weeks. Your comments, thoughts, and ideas are welcome.<br>
<br>- Kim Quirk<br>Dir of Tech Team<br><br><b>Power and Hardware:</b><br>The first Multi-Battery Charger sample parts off of the tooling made it to 1cc this week. Richard has been spending most of his time purchasing, scrounging up, and assembling all the other parts in order to create a working prototype. Once assembled it will start to get used at 1cc.<br>
<br>
Richard has modified his power logging script to try and do battery life estimation calculations as a first try for something that can be used<br>
for the battery monitor in sugar. Its available here: <a href="http://dev.laptop.org/%7Ersmith/olpc-bat_ttl" target="_blank">http://dev.laptop.org/~rsmith/olpc-bat_ttl</a><br><br><b>Tech Team in Deployments</b>:<br>
While in Peru over the past week, John Watlington queried individuals in the MED to prioritize software improvements being considered.<br>
Speeding up the user interface was high on everybody's list of improvements, with the proposed Sugar UI changes a close second. He showed them Joyride 1892 to illustrate the proposed changes, and discussed changes to the activity menu bar. Some of the technical team got excited and started downloading it to play with.<br>
<br>
A close third was improving the power management of the laptop to extend the life of a single battery charge. Following were improvements to the group model (specifically, they are looking at support of multiple groups, possibly defined and shared between users --- e.g. the teacher makes a group that is "first grade", so it is easy to invite only the first graders to join an activity.<br>
<br>
Improvements in the network association algorithm (the laptop needs to automatically identify school servers, irrespective of the connection method, mesh or AP, #6855) were assumed. We also need to continue the work on improving the reliability of the collaboration middleware under lossy network conditions. Several new patches have arrived from Collabora to test in the (in limbo) mesh testbed! Re-enabling support for automatic mesh portal is a topic of increasing importance. A very strong requirement is that further machines arrive in Peru from manufacturing with build 703 and the Peru activity bundle installed.<br>
<br>
Backup to the school server wasn't highly rated, perhaps not surprising given the sample set. Unfortunately, deployments haven't gotten to the point of obtaining feedback directly from the field on Update.1. (Arahuay just upgraded from B2 machines to mass production XOs and Update.1 two weeks ago.).<br>
<br><b>Hardware:</b><br>
A meeting was held at 1CC this week to kick-off the second generation of OLPC laptop hardware. The overall goals of this generation have been established for some time: 1) Much Lower Power, 2) Lower Cost, 3) More robust and easier to repair, and 4) Increased performance without sacrificing any more important goal. A surprising amount of consensus was reached on the overall form factor and look and feel, and more information will be publicly announced soon.<br>
<br><b>Software:</b><br>Scott Ananian:<br>* Most of my time was spent on networking issues. I wrote up a set of networking principles to guide deployment networking assumptions:<br>
<a href="http://wiki.laptop.org/go/Network_principles" target="_blank">http://wiki.laptop.org/go/Network_principles</a>. This document is still a work-in-progress; I'm working on integrating comments and prototyping some ideas to firm up proposals.<br>
* Towards the end of the week, I investigated the skype, Ubuntu Mobile, and Edubuntu ports to the XO.<br><br>Chris Ball:<br>
* Working on a manual "extreme low-power" mode. Came up against obstacles one and two: the EC commands for turning off the wireles device by putting it into reset are not implemented in the driver, and we'll need to be able to create per-kernel initrds before we can make the USB controller modular. Started work on both of these.<br>
<br>
* Once these are done, we can add a manual UI switch that turns off the wireless chip and unloads the host controller (which will give us about 300ms resume time and large power saving) before enabling OHM's idle suspend mode and backlight dimming.<br>
<br>
* Realized that the "wikipedia-iphone" open-source project (which stores compressed wikipedia snapshots on the iPhone and allows browsing of the snapshot by decompressing articles on-the-fly) is a great fit for us. Worked with Mel Chua to document what the steps to port it to an XO activity would be. Please see Mel's write up: <a href="http://wiki.laptop.org/go/Wiki_server" target="_blank">http://wiki.laptop.org/go/Wiki_server</a>. The full text of the Spanish Wikipedia is 400M compressed; by selecting only popular, well-linked, regularly-edited articles, this could drop to a size where it makes sense to put it on XOs directly. 400M is already appropriate for putting on a school server.<br>
<br>Javier Cardona on the thin firmware project or soft AP:<br>- 8388 mac80211 driver and firmware are functional for infra and monitor mode.<br>- device can scan, associate and transfer data (for now with no security and at fixed rate).<br>
- b43 (broadcom) + mac80211 + hostapd AP solution has been tested with latest hostapd and wireless kernel + necessary patches. This will be the golden system that we'll compare our libertas based solution with.<br>- build and testing procedures have been documented.<br>
- we should be able to make a first development release in two weeks (it will not include AP functionality yet).<br><br>Morgan Collett:<br>* worked on <a href="http://wiki.laptop.org/go/Tubes_Tutorial" target="_blank">http://wiki.laptop.org/go/Tubes_Tutorial</a> to get it in line with changes to HelloMesh<br>
* worked on a proof of concept to launch Chat when a non-Sugar Jabber client initiates a connection (testing, almost ready for review)<br><br>Guillaume Desmottes:<br>
- Investigated few presence related bugs<br>
- Continue to implement Gadget client. Most of it is done now.<br><br>Ricardo Carrano:<br>* Finished testing firmware release 22.p10. It fixes many bugs (#4616,
#6709, #4927, #6589) and it will be released to joyride as soon as Javier Cardona rewrites/approves the necessary driver patches. <br>* Summarized a list of current wireless issues in <a href="http://wiki.laptop.org/go/Wireless_Issues_Apr08" target="_blank">http://wiki.laptop.org/go/Wireless_Issues_Apr08</a> and started the <a href="http://wiki.laptop.org/go/Network_Issues_May08" target="_blank">http://wiki.laptop.org/go/Network_Issues_May08</a>
to collect contributions for a new list with the more ambitious goal of
concentrating all network-related issues. Please contribute!<br><br>Andres Salomon:<br>- worked on getting more stuff upstream.<br>
- merged master up w/ 2.6.25, discussed branching a new stable from it<br>
- created and merged newmaster up w/ linus<br>
- fixed a number of bugs in master and newmaster<br>
Linus's 2.6.26 will be the first upstream kernel capable of booting on an XO.<br><br>Michael Stone:<br>
This week, I reviewed touchpad bugs and prepared some straw-man text around which to organize further development. I also helped Dennis<br>
experiment with Sugar-on-F-9-on-XO with Dennis, critiqued several patches by Martin Dengler, critiqued Eben's Journal redesign work<br>
with Ben Schwartz and Jameson Quinn, and reviewed parts of Scott's "Network Principles" document with Daf.<br><br><b>Support:<br>
</b>Kim Quirk and Adam Holt held a G1G1 post mortem review to capture
the important things that were done right and done wrong. We will
deliver a summary of this to all interested people at OLPC in another
week. Attending the meeting were representatives from all our partners.
Some highlights: It is critical to appoint someone with oversight and
accountability for the entire program as too many things fell into
holes with no owner. Also most people felt that they 'pulled out all
the stops' for this program to the point of burning bridges and burning
out. Every partner ended up getting involved in the time-consuming and
stressful job of supporting people who were very upset. <br>
<br>
Emily Smith:<br>
<div>- sent out 50+ tax receipts for donated XOs</div>
<div>- continued to work on shipping problems for G1G1</div>
<div>- continued to send out replacement machines for RMAs</div>
<div>- Began manually notifying RMA recipients of their FedEx tracking numbers <br>
</div>
- Found that about 10% of RMAs reshipped have problems with delivery (address change, requested refund not reshipment, etc)<br>
<br><b>SysAdmin</b>: <br>
Henry Hardy:<br>
* All monitored systems had 100% uptime.<br>
* Crank (dev, trac, git)
continues to be slow. On April 28, a small problem caused by an
apparent hash collision between 2 tickets which contributed to added
slowness of trac occured and was resolved.<br>
* Upgrade of grinch filer is scheduled for this upcoming weekend. We
plan to add nine 1 terabyte disks to the coraid 1521 filer. Grinch will
be unavailable from Saturday at noon EDT Saturday May 3 to Monday May 5
at 09:00 EDT.<br>
<br>