Ideas: Google Summer of Code 2013

Status: Proposed - Draft

Summary of idea: GTK+ (GIMP Toolkit) is the cross-platform widget toolkit used in MATE to create graphical user interfaces. GTK3, released in 2011, and in Fedora since F14, brings a lot of improvements. Add GTK3 support to MATE, keeping the traditional desktop metaphor, is one of the goals of the MATE Desktop project.

Knowledge prerequisite: C, GTK.

Skill level: Medium

Contacts: Stefano Karapetsas

Mentor(s): Stefano Karapetsas, stevezesch2 [AT] gmail [DOT] com

Notes: -

Status: Proposed - Draft

Summary of idea: Wayland is a next generation display server protocol with the aim to replace the old X server. With Wayland, the window managers will become display servers and compositors at same time, using the Wayland library. Wayland is already available in Fedora, and KDE and GNOME already are moving in this direction, so this could be a great improvement for both MATE and Fedora.

Knowledge prerequisite: C, X11.

Skill level: High

Contacts: Stefano Karapetsas

Mentor(s): Stefano Karapetsas, stevezesch2 [AT] gmail [DOT] com

Notes: -

Status: Proposed

Summary of idea: The install script is one of the very first things a new Geeklog user encounters. Therefore, it is important that it is easy to use, supports the user in installing, updating and migrating a Geeklog site while being robust and providing helpful hints and error messages.

The current install script was originally implemented as a GSoC project in 2007 and further improved upon in GSoC 2008. We are still pretty happy with it, but there’s always room for improvement. And this is what this GSoC project is about - we are looking for an evolution, not a revolution.

Knowledge prerequisite: PHP, MySQL (possibly also PostgreSQL and MS SQL), basic webserver knowledge

Skill level: Low - Medium

Contacts: Dirk Haun

Mentor(s): Dirk Haun

Notes: http://wiki.geeklog.net/index.php/SoC_install_script_refinements

Status: Proposed

Summary of idea: Geeklog has been translated into over 30 languages, but many of those translations haven’t been updated in a while. The current way of editing a PHP file is just too awkward, error prone, and not very intuitive (since you can’t see the context in which a text string is used).

We’re looking for a way to make translating Geeklog easier, especially for people who do not have a knowledge of PHP. Also, for new translations, the sheer amount of text to translate can be discouraging. The idea is to address both of these problems by crowdsourcing.

Knowledge prerequisite: PHP, SQL, JavaScript

Skill level: Medium - High

Contacts: Dirk Haun

Mentor(s): Dirk Haun

Notes: http://wiki.geeklog.net/index.php/SoC_crowdsourcing_translations

Status: Proposed

Summary of idea: Geeklog currently uses custom session management to maintain users session information. The custom session implementation is antique and needs to be upgraded to a fully supported session management system such as the one provided by PHP Sessions. The implementation needs to be secure and should cache user data between page loads of a session.

Given Geeklog’s focus on security, this implementation must be hardened against the usual attacks.

Knowledge prerequisite: PHP, PHP / Web Security

Skill level: Medium - High

Contacts: Dirk Haun

Mentor(s): Dirk Haun

Notes: http://wiki.geeklog.net/index.php/SoC_php_sessions

Status: Proposed

Summary of idea: The system should use an appropriate networking profile (e.g. Proxy settings) for each network connection.

Gnome 2 had a concept of network locations in its Network Proxy configuration window. However, user should selected the appropriate location whenever he moves between networks. This idea is about providing an integration between NetworkManager and Desktop environments so that a user can create network profiles for each network location(connection) providing appropriate settings like proxy settings which is the main proposed setting here. NetworkManager can have a "Network Location" concept: for wireless networks, usually the name of the network (ESSID) is usually enough. For wired connections, DHCP servers can and usually do provide network’s domain name, which can be used as the name of the location. It is nice if a user can associate each network location with a network settings profile which will be used whenever the user is connected to that network automatically. So, when you connect to a network, a corresponding network settings profile is activated automatically.

Knowledge prerequisite:

Skill level:

Contacts: Hedayat Vatankhah

Mentor(s):

Notes: There is an entry with some description in NetworkManager TODO which should be considered for the implementation

Status: Proposed - Draft

Summary of idea: develop a free alternative of cpanel / plesk control panels, 100% compatible with fedora, and redhat enterprise Linux. written in python.

the control panel will be able to create domains and automatically setup apache, postfix, dovecot, mysql, postgresql bind etc…​

Knowledge prerequisite: apache, postfix, dovecot, mysql, postgresql, proftpd, bind

Skill level: Medium

Contacts: itamarjp [AT] fedoraproject [DOT] org, kaustubh [DOT] karkare [AT] gmail [DOT] com

Mentor(s):

Notes: webhosting control panel draft

Status: Proposed - Draft - Need a mentor

Summary of idea: Creating screencasts on Fedora and post-processing them for upload should be easy and fool proof.

The Fedora Videos project has been launched to bring together a collection of screencasts that help people learn how to use Fedora. There’s only one major hurdle, though. Capturing really good screencasts on Fedora, post-processing them to include an intro and outro and other effects, and finally exporting them to an open format is challenging. There are lots of different tools, but each one only gets you part of the way there and you have to be the glue to make it all work. And then there are the bugs.

The goal of this project is to assemble a toolchain (a recommended collection of software available in the Fedora repositories) and write a comprehensive tutorial for how to create an professional-looking screencast. That may included chasing down & fixing bugs in the existing tools, discovering new tools and getting them packaged and learning about techniques in the environment around the computer, such as how to select & setup a microphone or how to draft a good speaking script. One possible task in the project is to add sound recording support to the Gnome 3 desktop recorder. Currently, the desktop recorder only grabs the video of the screen.

The student will get support and advice from the Fedora Videos team since they are trying to learn how to create these screencasts.

Knowledge prerequisite: gstreamer, video and sound editing, blender

Skill level: Medium

Contacts: Fedora Videos Team

Mentor(s): -

Notes:

Status: Proposed - Draft

Summary of idea: Improving Fedora’s Tablet User Experience.

The window managers/desktop environments officially supported by Fedora do not match the expectations of people with low power touch devices.

Fedora features new technology to reduce system load and - requirements, the user interface has not seen too much love.

Currently available interfaces for users are:

The E17 Desktop Shell, developed as part of the Enlightenment Project, is a perfect match. It provides a rich interface, whilst consuming very little ressources. The Enlightenment Foundation Libraries, on which E17 is built upon, were chosen as Foundation Block of Tizen, the MeeGo successor backed by Samsung and Intel. If things work out, E17 will be the first "real" Desktop Shell for Wayland.

Though it already provides special behaviour for touchscreen devices through certain modules, there’s still room for improvement.

The overall goal of this project would be to improve the user experience of people who want to use Fedora on low power devices such as tablets.

The student will get support and advice from the Enlightenment Project developers.

Knowledge prerequisite: C, X11, Wayland, Enlightenment Foundation Libraries

Skill level: Medium to High

Contacts: Michael Bouchaud (yoz [at] efl [dot] so), Leif Middelschulte (leif [dot] middelschulte [at] student [dot] kit [dot] edu)

Mentor(s): Michael Bouchaud (yoz [at] efl [dot] so)

Notes: If this idea gets realized, a proper tablet spin can be created, that honors the spirit of Fedora.

Status: Proposed - Draft

Summary of idea: In transporting data across networks, programmers need GlusterFS language bindings to be able to create apps in their language of choice. A language binding translator would greatly increase the number of programmers who would be able to more easily extend GlusterFS. It’s also a fun project for someone wanting to get some experience with multi-threaded architecture and design in the context of distributed systems and data.

Knowledge prerequisite: C, Python, embedded Python

Skill level: Medium to High

Contacts: johnmark [AT]redhat [DOT] com , abperiasamy [AT] gmail [DOT] com

Mentor(s): Anand Avati - Mentor, AB Periasamy - Co-mentor, John Mark Walker - Co-mentor

Notes: -

Status: Proposed - Draft

Summary of idea: In general, there is a need for the ability to store and access NoSQL data on general purpose distributed filesystems. Currently, there are a host of single-purpose methods for storing and retrieving NoSQL data, which are difficult to access from legacy applications. Creating a NoSQL translator for GlusterFS would help mitigate this problem and give developers more options for storing and accessing "big data" in a way that is accessible via a variety of standard toolkits and protocols.

Knowledge prerequisite: C, GlusterFS internals

Skill level: High

Contacts: johnmark [AT]redhat [DOT] com , abperiasamy [AT] gmail [DOT] com

Mentor(s): Anand Avati - Mentor, AB Periasamy - Co-mentor, John Mark Walker - Co-mentor

Notes: -

Status: Proposed - Students' Idea - Need a mentor

Summary of idea: Today, many developers are working with different DBMS. I propose to develop a common command line interface for most popular databases. For example: MySQL, Oracle Database, PostgreSQL.

Knowledge prerequisite: Java, JDBC, MySQL, Oracle Database, PostgreSQL.

Skill level: Medium

Contacts: Alexandr Egorov 0x0j023 [AT] gmail [DOT] com

Mentor(s):

Notes:

Status: Dropped — This is a GSoC 2012 idea and is being working on by Hongyi Jia although he didn’t get the GSoC slot (http://www.jiayi.net/works/nfsfscache-writeback-cache/). So, I’m not going to mentor the idea this year and if anyone is really interested in it, feel free to contact me and I’ll be happy to answer questions:-)

Summary of idea: Currently, the flashcache/bcache only works for local filesystems. For network filesystems, nfs only supports write through cache based on fscache/cachefiles. With wider adoption of SSD on nfs client side, it would be great for nfs to support writeback cache to speed up write intensive clients. This project would implement a writeback mechanism for nfs, which need to make necessary changes to kernel nfs, fscache and cachefiles modules, and take care of nfs close-to-open semantics.

Knowledge prerequisite: C, Linux kernel, nfs

Skill level: High

Contacts: bergwolf [AT] gmail [DOT] com

Mentor(s): Peng Tao - Mentor

Status: Proposed - Draft

Summary of idea: Free software designers don’t have a great set of tools to work with to collaborate with each other and with the community on their design work. With this project, we’d like to make progress towards fixing that. We have some disparate ideas / tools that we’d like to be integrated:

Knowledge prerequisite: Ruby on Rails, web development, some UI design chops Skill level: Intermediate Contacts: edirsh at redhat dot com

Mentor(s): Emily Dirsh

Notes: We can provide a development platform for the web UI at openshift.redhat.com. Also, note this project doesn’t just benefit Fedora designers, but it benefits all FLOSS designers.

Status: Proposed

Summary of idea: Unit tests are good. Testing is good. fedpkg and it’s backend rpkg is a growing code base, gaining more and more contributors. It lacks a unit test framework which would be very beneficial to the code base for making sure contributions don’t break the code in subtle ways.

Knowledge prerequisite:

Skill level:

Contacts:

Mentor(s):

Notes: This project is in python.

Status: Proposed

Summary of idea: To package GitLab for Fedora and EPEL (if possible).

Git has fundamentally improved the way that developers share code. The barrier to sharing code has virtually been eliminated. We are also seeing the emergence of a new dynamic called "Social Coding". There’s no better example of this than Github.

Creating an environment to foster social coding, and a low barrier to sharing code, requires more than just git, though. A lot of what makes Github successful with git is the web-based front end.

Fortunately, there is an open source application named Gitlab that provides much of the interactive functionality and usability that is found at Github. The goal of this project is to bring that experience to fedorahosted.org by setting up Gitlab.

This project will require working with ruby packaging sig to get the necessary gems packaged and available in Fedora (and if possible EPEL as well).

Knowledge prerequisite: Ruby, git, packaging

Skill level: Medium

Contacts: Dan Allen (general), Vít Ondruch (packaging)

Mentor(s): Ranjib Dey (tentative), Dmitriy Zaporozhets (dzaporozhets), Ariejan de Vroom (ariejan)

Notes: This idea is definitely something the Fedora project is interested in pursuing. The work in this project will be carried on after the project, and is an opportunity for a longer-term involvement in Fedora.

A demo of Gitlab can be found here: http://demo.gitlabhq.com

Status: Proposed

Summary of idea: Contribute to Dorrie, and make it more usable, add test coverage and deployable. Details at Dorrie.

Knowledge prerequisite: Python, Django

Skill level: Intermidiate

Contacts: Shreyank Gupta

Mentor(s): Shreyank Gupta, Ratnadeep Debnath

Notes: Features that I am looking to target:

* User management/FAS integration
* Un-linerize the workflow.
* Default wallpaper/Hostname/user-password/root-password
* Custom repos/RPMs
* Home directory content
* Search Packages
* Image type: ISO, virt image, raw disk

Code at https://github.com/shreyankg/Dorrie, fork and send pull request, if you are interested to contribute.

other secondary architectures and distributions. Different workers should use different job queues and they all should work independently. The current admin commands also need proper upgrade to handle different resources to import from. The code also needs to be rebased so that it becomes more test friendly.

Status: Proposed - Students' Idea - Need a mentor

Summary of idea: The idea is to make an AUR clone for Fedora where users can submit buildfiles, which need to be hosted. These set of files can compile packages on the user side, and installed locally.

The benefit of this method is:

Contacts: Anuj More

Notes: Need mentor. Need advice on feasibility of the software and issues that can arise. Also, read the extensive blog post on my blog.

Analogy: This is a lot like "having your own library" vs "getting a book from a bookstore". If you refer a book often, you’d keep it in your library. This would occupy space and consume resources. Otherwise, you’d just pull out a separate paper with instructions about the bookstore to find the book, and fetch the book from the store. Though the first method is hassle-free, it is not possible to have all the books in the world with you.

Status: Proposed

Summary of idea: With the inception of Maven we have a means to standardize the building of Java projects. However it has also resulted in the usage of a very wide dependency set by these Java projects. This makes it very hard to integrate those projects on a single platform (be it Fedora or JBoss Application Server (or the combination)).

In essence during packaging all these dependencies must be lined up to form a single consistent component set. Thus each project will only have those components available which are actually available on the target platform. This is very hard to enforce using standard Maven setup.

For Fedora I’m envisioning a Maven extension that makes sure only sanctioned components are used during the build (without putting a burden on the packager). This will allow developers who are not on Fedora to also build with this extension and thus verify (and fix) issues which are the result of "Fedora packaging".

A prototype can be viewed at Maven FOSS Repository Extension.

Knowledge prerequisite: Knowledge of Maven

Skill level: Medium

Contacts: Carlo de Wolf, User:Mizdebsk[Mikolaj Izdebski]

Mentor(s): Carlo de Wolf, User:Mizdebsk[Mikolaj Izdebski]

Notes: You can find us at irc.freenode.net #fedora-java: wolfc and mizdebsk

Status: Proposed

Summary of Idea: Create an educational app introducing software from Fedora Robotics Suite

The Fedora Robotics SIG is creating a Robotics Suite consisting of many packages useful in robotics. We want to develop a demonstration application introducing new users step by step to core packages like Fawkes and Player/Stage. It would feature multiple game-like levels with increasing complexity. The general task would be to instruct the robot to fulfill a specific task in a simulation environment.

Knowledge prerequisite: Developing this requires a strong background in C++, a background in robotics is preferred but not necessary. You should be able to familiarize yourself with new software quickly. User visible parts will require GUI programming using Gtkmm.

Skill level: Medium to High

Contacts: Tim Niemueller

Mentor(s): Tim Niemueller

Status: Proposed

Summary of idea: Improve the current RHQ-PyAgent written in Python that interfaces RHQ with OpenLMI to pick up metrics that are provided by OpenLMI from Fedora or RHEL systems. The agent is talking to the RHQ server via the REST api and push metrics to RHQ etc. This agent does/will not implement the full functionality of the RHQ java agent. The previous work is already pretty good, but it is cumbersome to add new OpenLMI-enabled servers and the types of the data model. The work of this year should concentrate on automating that process - perhaps after providing a range of IP addresses to query for servers. Similar for resource types, the agent would talk to the server to see what types are available and then check with the OpenLMI provider if this supports the same.

While RHQ is written in Java, this project does not require any Java knowledge.

Knowledge prerequisite: Python, Linux system administration, principles of REST

Skill level: Medium to High

Contacts: Heiko Rupp

Mentor(s): Heiko Rupp

Notes: RHQ wiki is at http://jboss.org/rhq, Previous years work: http://rhqagent.blogspot.com

Status: Proposed

Summary of idea: Fedora Easy Karma is a CLI application to help testers apply karma to installed updates and while it’s great at doing so, it doesn’t display much other information on what an update is supposed to fix or what can be done to test a particular update. A GUI tool that pulls in more information on packages under test would help in the testing process and hopefully help encourage less techical users to start testing packages.

Knowledge prerequisite: GUI tool kit (Qt or GTK), python

Skill level: intermediate

Contacts: Josef Skladanka (jskladan), Tim Flink (tflink)

Mentor(s): Josef Skladanka (jskladan), Tim Flink (tflink)

Notes: A longer description and some rough mockups can be found here

Status: Proposed

Summary of idea: The blockerbugs web application is used by QA to track bugs which are currently blocking release, proposed as blocking release or could be pulled into the release past a code freeze. While the app itself already exists, there are many features which we would like to see implemented including (but not limited to):

Knowledge prerequisite: Some python experience is required, the app is written with Flask so some experiehce with that would be helpful

Skill level: Intermediate

Contacts: Tim Flink (tflink)

Mentor(s): Tim Flink (tflink)

The image building project started last year as a gsoc project and while the progress was good, it isn’t quite to the point where we need it to be for everyday usage. The emphasis would be on getting local image building to work before more work on distributing image building tasks got started.

Possible project components:

Knowledge prerequisite: Some python experience is required, familiarity with Fedora’s ISO build tools would be preferred

Skill level: Intermediate to advanced

Contacts: Tim Flink (tflink), Amit Saha

Mentor(s): Tim Flink (tflink), Amit Saha