to my surprise, after a few minutes the board was actually full and it seemed that we’ll have a very interesting program ahead of us
met Zonker, the other openSUSE Mentor and Community Manager, told him I’m having a session on Sunday
found out that we’ll be attending completely different sessions :)
first session - ‘Monetizing Open Source without Being Evil’ by Mixxx guys
pretty good, learned a couple of real-life examples:
Firefox - has Google search on home page and in navigation toolbar (paid by Google)
Rhytmbox and Amarok - last.fm / Jamendo / Magnatune integration (increases program value and is paid by the studios/companies)
Ardour - optional donation 10$/month - has around 120 subscribers (1200$ a month)
short session about developing for iPhone by Matt from Adium; more particularly about how GPL and other open source licenses clash with iPhone SDK NDA
veeeeery long queue for lunch; agreed with Assaf Raman from Ogre3D that Americans love queues :)
everybody is throwing Google colorful frisbees and balls
lunch is actually tasty, but I miss warm food, every choice is a cold dish :(
going to session called ‘Human Supercomputer Workshop’ held by James from Audacity
have no idea what this could be, but sounds like fun
9 people divided into groups of 3; each group has worker, facilitator and observer
worker says how he/she would solve the problem and also refines the problem itself
facilitator can interrupt worker to prevent preconceptions of the speaker
observer notices when the interactions were most effective and is allowed to give feedback only at the end of the session
session lasts for around 15 minutes and after that the groups are dissolved and the new ones are created, ideally so that none of the members of the new group is in group with the person he has met before
after 3 or 4 sessions, one large group is created and everybody gives feedback
totally fascinated how this “architecture” could solve problems in a very short time
interesting thing is that human supercomputer is working better when it is “under-clocked”, i.e. when speakers speak more slowly, as compared to CPUs which work better if they are over-clocked
In my previous blogpost I described the project I mentored during this year’s Google Summer of Code event. At the end of October I was invited into Google Headquarters in Mountain View, California for Mentors Summit. Here you can find my remarks.
Google planned a dinner for us at Amarin Thai restaurant, left the hotel around quarter to six
used VTA public bus transport to get there; turned out to be pretty adventurous: one had to request the stop and the stops were not announced; fortunately had a map software in my cellphone so I was able to track my position by looking out of the window and advancing cursor in my mobile map :)
arrived to the restaurant; not so many people yet, soon started to change …
noted that the dependencies of the packages are the a big problem, Lance had to admit that I’m right; they actually use database of packages and their dependencies for each distribution, must be huge and hard to maintain!
This years Google Sumer of Code, student Peter Libič tried to implement an idea of Migrating Assistant. MacOSX contains utility which can import users, application settings and various files from old Macintosh to new one. Idea to port this functionality to Linux is not new. Something similar was created during GSoC two years ago by Ubuntu, but we tried to use different approach (object oriented C++) so the code is better extendable and maintainable.
Application support is not as wide as it probably should be, but because of the clean design, we hope that the number of supported applications will rise in the future. You can try Mango for yourself - packages are ready in BuildService and we are looking forward to feedback from you. See http://en.opensuse.org/Mango for more info and installation instructions …
I finally implemented very important feature of scout: It is now able to search for binaries in all enabled zypp repositories. SAT-solver files are used for this, so user does not have to install any external index files. I really would like to thank Klaus Kaempf for his exhaustive help with python bindings for satsolver. Another great news are that Werner Fink applied command-not-found patch for bash package, so 11.1 will probably contain this feature working out of the box! Current early implementation has one problem though: it is pretty slow comparing to older use-own-sqlite-database approach (2 seconds compared to 0.2 seconds). But it indexes more repositories at once (I have 14 enabled) and I believe the code could more optimized and thus whole search faster in the (near) future.
Feel free to test the packages from BuildService (follow the instructions on wiki) and tell me what you think of it ! :)
A lot has happened since the public release of scout. Blogpost registered more than 400 hits, Marek Stopka created bash-completion, Thomas Schraitle wrote docbook documentation and Michal Vyskocil prepared module for python and its indexes. Thank you all! I started a wikipage like Thomas suggested and indexed Packman repositories for their binaries.
Michal also prepared small demonstration video about using scout in java wrapper. The wrapper runs java application and greps stderr for exceptions. When NoClassDefFoundError/ClassNotFoundException is detected, the classname is taken to scout, which resolves it to package name, installs the package with zypper and tries to run application again! I like this idea pretty much. Michal is currently working on perl indexes and we will probably index also ruby and pkgconfig files.