Wanted: A GSoC student

Perhaps it's a bit early, but Toshio and i have been talking about a couple of crazy GSoC projects that will require a bit of a warmup cycle. I'm looking for a couple of highly motivated students who are looking for Summer Jobs and want to do something very crazy and awesome. If you're looking for something to do this summer, leave your email address in the comments, or use my contact information on the Fedora Wiki.

The first project is a content delivery project. Lately we've been bumping into issues with Fedora's No Content policy where people want to deliver wallpapers, themes, books, documentation, Gimp brushes, and so on that are essential to adding value to the packages we provide in Fedora. Alot of this content falls into a grey area that makes it really hard to decide if it should belong in Fedora. In most cases, however, it really doesn't belong there. Faced with the decision to turn down content submissions, we'd like to create a repository where we can point people to instead.

The aim of this project is to put together a content database system and some delivery mechanism and standard to get content to people's computers. What the database is going to look like, and what kind of frontend is completely open to the student working on the project. In the coming months, the goal is to sit down and discuss how you want to solve the problem, and how you plan to go about doing that. We'll plan out what standards already exist to solve part of the problem, who is using them, how to handle licensing questions, how to store data to make it friendly to the systems involved, and what's the best way to locally 'install' it to the computer. A project like this can become very complex, and the biggest challenge is refining it into a 12 week project.

The second project is a bit simpler. We want reliable download statistics from mirrors who mirror linux distributions in general. If you are familiar with Smolt, this will sound very familiar. We're looking for someone to write some tools mirror managers can use to gather statistics on their machines and a way for the stats to be gathered publicly. The details of the implementation are left to the student, but since we're looking for the right umbrella organisation and place to host the project, we are also recruiting now.

Both of these projects will most likely lean heavily on Fedora for support, since we plan on making heavy use of both of them once they are production ready. Both projects will be under a different organisation though, and you do not need to be a Fedora contributor to participate. The only requirement i have is a couple of years of experience with Linux.

5 flames:

Ant Bryan zei

for reliable download stats, you should partner up with MirrorBrain. they're working on that.


Bert zei

I think you know my e-mail address :-)
I vote for the first!

Anoniem zei

I was wondering if it would be a good idea to investigate whether EKG could be turned into a "generally useful, distribution agnostic tool". Well ! I really don't have any idea if EKG is strictly hard-coded to understand the Fedora side of things, but it does seem to be a prudent approach to make it modular with pluggable modules that can be used by various other projects.

Kevin Kofler zei

The current state of the art for content delivery is the opendesktop.org (not to be confused with freedesktop.org) network, also known as kde-look.org, kde-apps.org, gnome-look.org and several other subsites.

KDE uses its API to fetch content (data-only stuff such as wallpapers) and other stuff which does not need distro-specific packaging, such as plasmoids (Plasma applets/widgets) written in interpreted scripting languages. That feature is called GHNS (GetHotNewStuff).

OpenDesktop recently implemented an improved API called the OCS API 1.4 (see also the previous update, OCS API 1.3), which adds several new features, e.g. content uploads directly from the client application, without having to go through the web interface. The specification has been published on freedesktop.org.

A Qt-based implementation of the OCS API called attica is being developed together with and used by KDE 4.4. I see no reason why this API couldn't be implemented by other desktops.

The API is also not hardlocked to the opendesktop.org sites, other content repositories can also implement it, and KDE 4.4's implementation allows an application to offer several content sources to its users. So a Fedora-run content server could be envisioned for some applications.

Bert zei

btw, should I write you a motivation e-mail? :-)