VLC Dev Projects
This is a list of small independent improvements to VLC media player for which we are looking for interested developers, because we simply don't have enough time.
These do not require advanced knowledge of VLC internals but they can still vastly improve many parts of VLC
For each project, we list a contact person, who knows the subject best and who will be able to assist you if you are interested in working on these. We generally are very often present on irc.videolan.org, chan #videolan, so please don't hesitate to drop by or to send a mail if you are interested in a given problem (our mail adresses are generally <nick>@videolan.org).
Be careful, some projects may be unrelated, or deprecated, because of fast evolving software! Ask before !
- 1 Easy projects
- 2 Network
- 3 Inputs
- 4 Video
- 5 Interfaces
- 6 Playlist
- 7 Projects running
- 8 Finished projects
These projects are junior projects, the most easy ones to do, that require minimum knowledge of VLC.
We also have a number of "easy" bugs. You can check them out here
It basically means to clean the code, it is helpful for all programmers. It helps to find bugs, and is a good way to learn about VLC's internals.
It is covered in the Code Janitorial section.
if you feel confortable with english language, be sure to have a look to the QT4 Strings Review.
You can do video filters for VLC media player.
You can inspire yourself on this list of filters from a commercial product.
You can start the code here.
Talk to dionoea or funman
Performant VLC streaming server
There are many bottlenecks currently limiting the performance and scalability of VLC as a streaming server:
- The I/O blocks are linear, which implies lots of avoidable memory copying and dynamic memory allocations (particularly in the streaming output).
- mdate() (the timestamping function) is invoked in too many places to count them; but it is not a cheap operation on modern systems and should be avoided on the fast path. Ideally, a single thread would only need to call mdate() at most once per wake-up. In some case, mdate() is only used for debugging purpose which is really inefficient.
- The HTTP/RTSP core is single-threaded. This prevents scaling to SMP systems (which are the norm on server-side nowadays) properly. This also implies not very scalable I/O event polling. Finally, this gets problematic if one HTTP/RTSP client triggers a computationally intensive operation (such as establishing a TLS context if HTTP/SSL is used).
- There are many dynamic memory allocations that could be avoided. Video filters and codecs are very careful about this, but the stream output plugins are not. In many case, a stack-based buffer can be used (stack buffer, has constant time fast lock-less allocation, and needs not be freed explicitly) instead.
- The messaging subsystems might be adding more locking (needs checking); this should be avoided particularly if the message is anyway ignord (e.g. debug message in non-debug mode).
Other improvements that would be particularly benefitial on the server side:
- Memory allocation failure are not always duly handled. In some cases, that could potentially be leveraged by a client to crash a VLC server by causing a very large allocation attempt.
- Rewrite and revive the root wrapper to allow VLC to bind to the RTSP (554) port while not retaining root privileges.
Windows Media Center (WMC) IR Remote Control
Basic Controls work (e.g. Volume, Mute etc), but I would like to get Play/Pause/Stop/DVD Menu etc to work!
DirectX 9 video output
DirectX 9 features some improvements that we could use in our video output
Improvements to the X11 video output
Fullscreen support to improve .
Pixel shaders in OpenGL video output
It would be nice to have a small proof of concept video output which takes care of chroma conversion using pixel shaders.
This is a big subject, that has barely started. Interested persons will need to work closely with existing developers to implement this. Not much information is available yet. The idea is to make streaming very easy, and yet powerful by using parametrable profiles. These could be used either by the graphical interfaces or through HTTP/command line interfaces
VLC.app rewrite using the Mac OS X Framework
Improved Podcast support
Our current podcast support is very weak. We can read them, but that's all. We need a real infrastructure to handle them correctly. This will require some changes to the playlist code, so you will need to work closely with existing developers.
Improved ASX support
ASX is a playlist format used by Microsoft stuff. We can read it but we could do much better. More information is available at https://trac.videolan.org/vlc/ticket/20
We currently have some miscellaneous code to detect discs (DVDs, Audio CDs, ...) but it's a bit scattered and it could be used much better in the playlist.
Some programs support so-called live folders. You add a folder to the playlist, and if the files in the folder are changed, the changes are copied to the playlist.
Having this would of course be great :)
It might require some work to be able to do this in Linux, Windows and OS X, but the current developers will of course be able to help on these.
Hint MP4/MOV Files
Add video and audio hint tracks as the output file is muxed. Resulting files can be streamed from RTSP server with trickplay, like DSS/QTSS. Probably requires use of GPAC libraries.
Plugin is finished, now part of vlc 0.9.0 (development version). Done by funman
DBus is a communication mechanism for Linux (mainly).
We would like VLC to be able to use it so that other programs can interact with vlc (start/stop, know what is being played, ...)
Work is well advanced, see DBus, by funman and Mr_Mirsal
Our RTSP implementation (for broadcasting and video on demand) does not support the RTCP protocol. This is a must to have a real support for RTSP. Some work has already been started, but there is still much to do.