[OLPC-devel] Summer of Code introduction, progress update

[Ivan Krstic]

This year, OLPC is one of the mentoring organizations in the Google
Summer of Code program. We received more than 75 applications, and had a
hard time choosing only 5 to mentor. Congratulations again to the
selected students!

This is the introductory SoC progress update, and the only one I'll be
compiling. I've asked the students to send progress updates to the
devel@ list every Sunday. I hope to see some great work from this bunch.


o   Matthew Harrison is a freshman studying computer science and
    astronomy at the University of Massachusetts Amherst. He's working
    on external data sources (EDS) for our eBook reader. This means
    things like importing books from Project Gutenberg, the
    International Children's Library (ICDL), and so on.

    Matthew reports: """I've got code to "detect" all the modules in a
    modules folder, code to download and save the PG catalog locally,
    code that strips the catalog of things that aren't book listings,
    and I've outlined the catalog search functions.  This week I'm going
    to work on finding the best/fastest way to search the catalog and
    return results.  I want it to be able to search by title, author,
    keyword, language, and etext number.  As for searching by language,
    PG appends "[Language: Foo]" to books that aren't in English on the
    next line after the book listing in the catalog, so it shouldn't be
    too hard to do language searches.

    I'm still looking into finding the best format for the books.  I'm
    still leaning towards XML or something XML-based, but depending on
    how exactly the interface is going to be implemented, some formats
    may be easier to work with than others."""


o   Eric Astor is a freshman at Swarthmore College, with an intended
    Math/Physics major. He is working on building a set of tools that we
    can use to do two things: one, to take a dump of Wikipedia page
    (such as the Simple English Wikipedia, or any other one) and produce
    an automatic ranking of the pages therein based on "importance",
    which really means graph-theoretical network centrality. This lets
    us say "what are likely the most useful pages to have if we can only
    dedicate 25MB to having encyclopedia content on the laptop?".
    Presumably, the list is then fed to human editors who make the final
    decisions, and processing this final list is the second part of
    Eric's work; his tools are to take it and produce from it a
    locally-browseable Wikipedia dump.

    Eric recently ran the Simple English Wikipedia through a
    PageRank-like algorithm with the help of an open source graph theory
    library, and reports: """As it turns out, the results from the
    PageRank run against the Simple English Wikipedia are very
    reasonably sized in OpenDocument format - so I've attached that.
    I've noticed a definite bias in the results - towards geography, and
    English-speaking countries in particular. I suspect that this is
    mainly because the group of contributors to the SEW is small and
    specialized."""

    I've posted Eric's result set here:
    http://solarsail.hcs.harvard.edu/~krstic/simplewiki-pagerank.ods


o   Eduardo Silva, from Chile, is studying networking at Inacap
    (http://www.inacap.cl) and is working on a new memory usage tool for
    developers.  He's trying to unite various sources of information
    about memory consumption, and present it in a simple, useful numeric
    and graphical form. His new tool, which we dubbed Memphis, will also
    support scripting with Python to allow for memory and CPU usage
    triggers and similar extensions.

    Eduardo reports: """The last 2 weeks I was reading and learning all
    about Python , GTK+ and Cairo.

    - The first step was to write a class in order to read procfs.
    - Write methods to get basic information for a given PID.
    - Design first GUI with Glade
    - Write methods in order to use our GTK TreeView where process
      information will be showed

    At this time, Memphis can:
    - read procfs and load data into a treeview: PID, PPID, program name
      and RSS
    - print memory status in console for a selected process in the
      treeview using just a static frequency

    Now I'm writing some classes to draw the stats in the graph for a
    selected PID."""

    The first screenshot of Memphis, still lacking the graph display,
    is here: http://solarsail.hcs.harvard.edu/~krstic/memphis-01.png


o   Arthur Wolf is working on a project that's potentially useful to
    OLPC in the long term: he's building a peer-to-peer solution for DNS
    and DHCP on mesh networks, to simplify deployment and use of citizen
    networks (free Metro Area Networks). He reports having started on
    the main P2P part, which he deems the hardest piece to do.


o   Erik Pukinskis is a graduate student of Human-Computer Interaction
    Design at the School of Informatics at Indiana University. He is
    actually a SoC student for the AbiWord project, but he's working on
    an AbiWord interface for OLPC, which means we get to lay claim to
    him for the purposes of this report. Erik reports:

    """A lot of what I've been doing so far is finding the people I need
    to be coordinating with.  Since there's a strong design component to
    my project, is I've made contact with a bunch of people who know
    some of the regions the OLPC is going to (Brazil, Nigeria, Thailand)
    and I'm interviewing some of them about what education and life is
    like for kids there.  I'm blogging about this here:

     http://www.snowedin.net/summerofcode/

    My first interview should be posted today. I've also been talking
    with Chris Blizzard to try to figure out what they have already
    decided about the UI so I can work within their expectations.  And
    I've got AbiWord building on my laptop, but I haven't started
    hacking much.  So that's basically where I'm at.  My next steps are
    to post on my interviews/user research, and what the emerging UI
    requirements look like, start sketching UI, and start playing with
    the AbiWord UI to learn how the codebase works.  I also want to get
    the SDK installed on this machine, but I am waiting for a new hard
    drive to come next week."""


o   Manu Cornet is a first year PhD student in bioinformatics at the
    École normale supérieure in France. He's done great work for Ubuntu
    during last year's Summer of Code program, and is now working on
    hacking GTK+ for our purposes. This includes minor tweaks (such as
    disabling cursor blinking for extra power-saving on the display) and
    major work on an OLPC-friendly GTK theme. He is being mentored
    by the inimitable Federico Mena Quintero of GNOME fame.

    Manu didn't send a progress report. Calls to his home phone,
    university phone, and cell phone have been unanswered. He hasn't
    been seen in days. Witnesses say he was last spotted in #olpc,
    where, we learn unofficially, his mentor chided unidentified
    "corporate bastards" for "doing his poor student's work". A full
    investigation is pending, with retribution by the corporate
    bastards being a possible cause of disappearance. We remain
    hopeful Manu is unharmed.

-- 
Ivan Krstic <krstic at fas.harvard.edu> | GPG: 0x147C722D