Summer of Code/2006

From OLPC
< Summer of Code
Revision as of 17:29, 18 March 2007 by Sj (talk | contribs) (soc archive)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

2006 projects

Directions

Mentoring

If you are interested in mentoring participants, please read the mentor FAQ. Note that both the ability to spend significant time mentoring, and the quality of the students are key to a good return on investment by the mentor, and the learning by the students. Please do not sign up to be a mentor unless you are certain you can meet the obligations of being a good mentor. If you are interested in mentoring, please send mail to Jg with a sketch of your background.

OLPC Summer of Code Projects

If you have any additional ideas, please send mail to Jg.

User stack architecture

We're looking for several exceptional students who are willing to spend their summer in Boston, working within the user stack team, building the architecture and the application suite which will be the children's primary mode of interaction with the laptop. This suite currently consists of the E-book reader, a wiki notebook, and e-mail and IM clients.

None of these are traditional applications like the ones we're used to; the OLPC environment poses unique and extremely difficult problems to solve. The development will be done in Python; for more details about what caliber of participant we're looking for, see the original internship announcement. If you're planning to apply for this position, please follow the instructions in the previous link and send a resume to Ivan before submitting a proposal through the SoC system.

In return for your outstanding skills, we're offering the ability to work on an incredible software project, on an unprecedented scale of deployment, with extremely bright people, and right out of OLPC's Cambridge offices. It'll be a wild ride -- jump in.

GTK+ Toolkit

There are a number of simple modifications needed in GTK+. Here are some examples: the cursor in a text area flashes indefinately, rather than turning itself off after a couple cycles. Similarly, we need a theme that is space efficient, works on both grayscale and color by careful selection of colors with differing shades of grey. And at 640x480 resolution, some tuning of the borders between widgets may be needed to enable fewer applications to require modifications to work at this lower effective resolution (note the frame buffer will always stay at the full resolution; turning to color mode reduces the effective resolution without changing the frame buffer size.

Power aware applets

We expect that a number of applets that display information will need modification to become better power aware.

Better memory usage tool(s)

Right now, ps, top, xrestop, and so on all give various, often misleading and confusing reporting of memory usage, that defies all but system wizards understanding. A utility that would gather up the various memory usage attributable to a process and report it in an understandable way for application programmers would help greatly in clarifying (mis)behaving applications and libraries.

Networking

Include mesh routing (such as http://www.olsr.org) to the OLPC laptop. Analyse and simulate scalability and performance issues. Contact: aaron@lo-res.org Analyse and agree on a power aware protocol or plugin. Must work with the Marvel chip.


Educational Content

HIV/AIDS Education

Create educational games to help children learn about HIV prevention and AIDS. We want content for children of all ages, simple to use and easy to understand. Collaboration with international AIDS organizations is recommended. Possible partners include the Nkosi Johnson AIDS Foundation in South Africa.

Proposer : Kpi 10:10, 2 May 2006 (CET)

One dictionary per child

Nutshell: Multilingual Wiktionary & Wikidata customization for the laptops/servers.

The Multilingual Wiktionary project is picking up steam, and will have some active functionality in a few months. A version of this [and of the underlying wikidata implementation] customized to work with the wiki-based reader software already on the laptops/servers should be produced. This would be combined with a parallel content project to produce a 100-language dictionary (with miminal swadesh/other word lists in a further 200 languages) that could be included with every server; and a subset of the same which could be included on each laptop.

Proposer : User:Sj 04:00, 2 May 2006 (EDT)

One encyclopedia per child

Nutshell: Wikipedia & Wikimedia Commons customization for laptops/servers

The idea of a "one encyclopedia per child" project to create a condensed snapshot of Wikipedia content suitable for all OLPC laptops (not just the servers) has been floated on this wiki, and on Wikipedia itself. There is an important software component involved -- the culling of statistics about what articles matter, and the scripted export of Wikipedia content (perhaps only part of each article) into a suitable format, both require work and consideration. On the statistics front, a variety of statistics should be gathered (by link-popularity, reader popularity, seminality in Wikipedia's internal-link network, centrality in its field, ...) and analyzed. On the scripting front, scripts should be developed to strip/modify external links; to pull out the first sentence, paragraph, or other subsets of longer articles; to properly reformat or remove image or media links; to properly handle templates and other wikitext features which may not be supported in the native reader.

As for handling of images and media, some work might also be done to better produce and store compressed thumbnails of content, so that a variety of such content may be included. Some form of temporary local caching of requested media, pulled from the nearest server, or of media sharing across machines on a mesh network, could also be useful.

Proposer : Dr. Ian Geoffrey Kennedy* (general idea), User:Sj (the specific proposal above) 04:00, 2 May 2006 (EDT)

* see User:olpcme, possibly Dr. Kennedy on this wiki


International Children's Digital Library (www.childrenslibrary.org) on The Laptop

The ICDL is a library for the world's children with books in 35 languages. Design goal is to make the books accessible on The Laptop. This work would involve considering file and data formats, the ebook reader, and image quality for The Laptop display.

Project #1: Develop an AJAX and/or Canvas web book reader that displays a layer of text as image on top of the book scan image so that the text image (which is one image per word) is resizable, reflowable, and searchable.

Project #2: Develop an AJAX and/or Canvas web book reader that moves the functionality of the ICDL spiral reader into a regulary web page (Example book in Spiral Reader)


Accessibility

Vision Test

In Brazil the number of people with a disability is 10%, and on others can be much more, with visual impairment corresponding to a big part of that number, so it might be a good idea to include a vision test software that can check the child's vision and sees if she needs to be pointed to a doctor to better use the Laptop.

Proposer: --Felipecarvalho 06:54, 2 May 2006 (EDT) (as a student)

Virtual Magnifying Glass

Nutshell: Improve the magnifier so it can take advantage of X Composite Extension and be much more adequate for the Laptop and much more efficient.


Virtual Magnifying Glass is a free, open source, multiplatform, screen magnification tool. It is simple, customizable, and easy-to-use.

The propose consists in adding Dynamic Mode to the software by using the X Composite Extension. This allows the user to see the current desktop contents under the magnifier window. The user can also interact with the window under the magnifier (such as mouse, scroll wheel, keyboard, etc...).

Proposer: --Felipecarvalho 07:12, 2 May 2006 (EDT)

twexter language learning interface

Nutshell: text formatter for wide world of language learners

http://twext.cc/dev/summer-of-code.html proposes open software enabling translators to easily format bilingual text to help kids learn languages. Interface enables users to easily create "twext" translations and to add new languages. System promises to build bridges between a wide variety of languages; especially useful with songs, pithy quotes, proverbs, sayings, poems, etc.. System could integrate with variety of softwares, including multilingual Wikis.. http://twext.cc/license is currently GPL/CCL, with flexibility to host variable licenses by permission. Twext language learning system subscribes neither to theory of grammar study (boring), nor to philosophy of "immersion" (stressful), but rather to Stephen Krashen's theory of "Comprehensible Input" http://www.sk.com.br/sk-krash.html

Proposer : [C] [1] duke crawford; mentor needed for this project to be further considered.


Rejected OLPC Summer of Code Projects

The proposed projects below were reviewed and rejected.


Indication of a font not having a glyph for a requested character

The following proposal is, for the reasons in the comments appended below, not useful

I have no knowledge of linux, most of my experience with PCs is with Windows, mostly Windows 98. So, I could not carry out this project, yet it could be something worth doing so maybe it would be a good project for someone else. I have no idea how big a job it would be.

I will try to explain what I mean in terms of what I am used to within Windows in the hope that someone who knows about both Windows and linux will assess the idea for possible benefit to the One Laptop per Child linux environment software packages.

Suppose that, in Windows 98, one opens a Unicode text file in WordPad. WordPad will use a TrueType font to display the text. If the Unicode text file has within it a character for which the particular TrueType font being used does not have a glyph, then the .notdef glyph of the font would be displayed. This could be an outline of a rectangle or some other shape as decided by the designer of the particular font.

However, if one is, say, opening a many page document, perhaps with a view to printing it, it may not be easily apparent if there are only one or two places in the document where the .notdef glyph is displayed. Even if there are no places where a .notdef glyph is displayed it could take quite a time to discover that that is the case.

My suggestion is that there be a facility within the software packages, a module which all such packages could use, that when a text file is opened that that module checks which character display requests to the font have a .notdef glyph returned for display and then either states on a dialogue panel that all characters have glyphs supplied by the font, or states on a dialogue panel that not all characters have glyphs supplied by the font and lists the Unicode codepoint of each such character and where it occurs in the document, which could be one or more places, also providing the option to save that information to a file.

As I mentioned above, I know nothing of linux. Does linux use TrueType fonts? Anyway, I hope that this suggestion is of interest.

Suggested by William Overington 2006-05-04 1104Z

Linux (the graphical environments like GNOME and KDE) do use TrueType fonts.

Linux does better than Windows, as it can show the codepoint's number in a rectangle for the missing character. For example, suppose that A (0x0041) is missing. It would show as a single character resembling

|---|
|0 0|
|4 1|
-----

Works also for Plane 1. In general, we should make sure that the supported fonts are installed already on the system. Linux (fontconfig) makes sure that if there is a font that has a needed glyph, it will use it.

Simos Xenitellis 134.83.1.225 16:50, 4 May 2006 (EDT)