Projects

Stuff I've worked on

To get at any of this, you'll have to be able to find my Arch archive. It lives at http://jeffrey.yasskin.info/archives/2003/. Once you've downloaded tla, use tla register-archive http://jeffrey.yasskin.info/archives/2003/ to tell tla about it.

Rhythmbox, an iTunes clone.: I made the random play order actually take into account your song ratings. Actually, the main contribution I made was pulling play orders out into a limited plugin-type system, but you don't really care about the details. My contributions live in the arch://jyasskin@mail.utexas.edu--2003/rhythmbox category.
GStreamer, a media framework.: I'm gradually writing a tag-printing element. It currently lives at arch://jyasskin@mail.utexas.edu--2003/gstformattags--main--0.8. As of gstreamer 0.8.1, it actually worked, though without a test suite. Since they're now on 0.8.5, I'm not very confident.

Stuff I plan to work on

These may never actually get done, but they're here. The ideas are listed here partially to provide prior art against any patent claims.

A content management system for this website that will let readers double click anywhere on a page and leave annotations or comments. Anyone should be able to respond to anyone else's comment. I should be able to delete arbitrary comments. The system should provide access to old versions of any page and diffs between them.
XSLT-based XML parser. The only two XML parsing interfaces I know of are SAX and DOM. SAX requires a lot of coding but runs quickly, while DOM requires less coding but more memory to store the tree. Since XSLT is designed to transform XML declaratively, I think some small tweaks could allow it to return an object tree rather than an XML tree. For simple queries and a good XSLT compiler, this could be as efficient as SAX. Yet it has very nearly the same power as the DOM for parsing. However, I haven't thought out the issues yet.
A simple C++ scripting library. I want to make C++ as easy to write scripts in as bash, perl, or python. The Boost.Filesystem library gets us partway there, but there's still no easy way to call other programs on the command line and redirect their input/output.
I'd like to re-do this website in RDF+XSLT+RDF Twig. It's already in Website (DocBook) + XSLT, but I think semantics are better than docbook…. Anyway, RDF Twig isn't necessarily the best way to transform RDF into HTML so this may become a big development project. The first thing I'm working on is my resume, which was in an ad-hoc format. I tried to convert it to XML Résumé, but they're missing a lot of elements, and I figure that representing corporations is best left to a non-resume standard. RDF allows me to punt on the things I want and define the things I know about, so I'm trying to use that.