New OLPC/Sugar Software: What's Next? & XO/software/hardware Q&A

1. Presenters
   1. Daniel Drake
      1. Did an internship at the OLPC offices doing software development for 8.2
      2. Worked on a deployment in Ethiopia followed by 7 countries in total, mostly in Latin America. Mostly working on technical training and logistics.
      3. Was on the team developing the XO1.75 laptop
      4. Involved in developing the next generation of releases for the XO1 and XO1.5
      5. A major goal is to make software development for the XO more sustainable to make it easier to keep refreshing the software
      6. Currently there is a lot of work when a new release of fedora is available
      7. Started to rebase the latest version of Sugar from Fedora 11, the current official version used by OLPC, onto Fedora 14. Its early days; an image is published but there are still lots of teething problems. Bug reports will be accepted later once the build is more stable
      8. There are 80 kernel patches outside of the official linux kernel. Daniel is through about 30 of them, others no longer need to be applied. Hoping to get the testing community up and running again over the next few weeks.
      9. The focus of the work is not to do a new release but to improve sustainability of the process.
   2. Stephen Parish, smparrish@sugarlabs.org
      1. Been working on Fedora for 4 years.
      2. Got involved with OLPC doing the builds keeping the 1.0 and 1.5 in sync
   3. Bernie Innocenti, Bernie@sugarlabs.org
2. New features
   1. Mesh v adhoc wifi support in XO1 and XO1.5
      1. The XO1.5 does not have mesh support; it has ad-hoc wifi network support in the same way as standard windows laptops do; this has improved features over mesh
      2. In the mesh all machines say they are here ten times a second on a clock beat; new machines joining the mesh synchronise to say “I’m here” at the same time
      3. In an adhoc wifi network one machine is the owner of the network and transmits ten times a second. If the king dies other machines try and take over. It works for small groups close together, like three kids under a tree. It does not work well for more widely dispersed machines with more participants. The adhoc network is also very low performance, much worse than the mesh.
      4. The XO1 supported mesh in hardware, even when they were in suspend node. The XO1.5 is moving to having mesh implemented in software, and also the machines are being designed to suspend more, even with the screen active, to increase battery power. This means the mesh will be less reliable with the 1.5
   2. Journal updates
      1. Shows file size and allows sort by this so kids can delete large journal entries to make space. Option discussed was to show the icon in a logarithmically scaled size so it’s obvious visually which entries are bigger
      2. Two click backup of the journal (using rsync). The restore replaces the entire journal with the backup. Backup could be done on one XO and restored to another one. Options considered for future:
         1. Check the date stamp on the journal and warn that changes since the date of the backup will be lost if the restore is run
         2. Considered implementing partial backup and restore to overcome these problems, listing the content of the backup to be able to drag and drop individual files
      3. Ability to repair the journal (which currently corrupts occasionally because of bugs in sugar and interplay between other software components). The repair rebuilds the journal index.
      4. Some bugs in dextrose fixed related to cleaning up data left hanging after journal entries or activities have been deleted such as the cache of the browser, or files used by wine, or files linked to from an activity or journal entry
   3. More intuitive resource monitoring added to provide visibility of memory and cpu usage. This was added partly to provide transparency to the problem that children keep starting new applications without shutting the others down and the machines run out of resources, which leads the teachers to complain that the software is not reliable.
   4. Considering adding a feature to automatically close the older running activities to free resources for the new activity the child has just opened. Other options being considered include preventing opening of new activities until others are closed if resources are low. This is difficult to predict as the system does not know what resources will be consumed by the newly started activity until it is running and starting to consume it, requiring management the resource allocation on the fly.
   5. Fast pen-tablet / touch-pad switching is being considered. There are known problems with using the full length of the touchpad in the field such as the mouse jumping around unexpectedly. They are hoping that some new ways of using the full size of the touchpad can reduce these problems.
   6. The browser now has tabbed browsing. This is classroom ready, having been tested for two months by 5th and 6th grade students in Paraguay.
3. Challenges
   1. Many children do not register their XO. The XO gives them no real feedback that the registration has happened and they cannot see any benefit so they don’t do it.
   2. Many children keep opening new activities without closing those they have finished with resulting in the machine running out of resources