Development discussions and questions take place on the Xiph.Org Opus mailing list (firstname.lastname@example.org).
Discussions related to the IETF process happen on the IETF codec working group mailing list (email@example.com).
For archives of recent discussions, try:
Informal development chat and support happens in #opus on irc.freenode.net. You can join the chat room through a web interface if you don't have an IRC client.
How To Contribute
There are many ways to contribute to Opus development:
- Reporting and fixing bugs
- Improving tools
- Improving testing framework
- Optimizations (assembly/intrinsics)
- Encoding quality improvements - see the Opus tuning page for suggestions
- Mapping to new containers - see MatroskaOpus, Mp4Opus and OggOpus (on Standards Track as RFC 7845)
It is generally advisable to contact the developers on the mailing list or the IRC channel before taking on new work on Opus, to avoid duplicating work and to make sure you're doing things the right way from the start.
How to report bugs
For sensitive (security-related) bugs, please contact the developers directly.
How to submit a patch
If you have a patch you would like to contribute, just send it to the mailing list. We can also take Github pull requests, but please send a note to the mailing list since the GitHub Opus repository is only a mirror.
Opus is the result of merging three different codebases and therefore does not have a consistent coding style. For example, the SILK code uses 4-space indentation, while the rest of the code is mostly 3-space, except the entropy coder, which is 2-space.
The general rule is that you should follow the style of the code you're modifying.
- Do not reformat the code in your "functional change" patches, as it mostly makes it harder to review changes.
- Do send separate "format-fixing" patches, making sure they don't change Opus' functionality at all.
For any new feature, it is strongly suggested to also include tests for the new code to make sure it works and keeps working in the future.
Opus only requires a C89 compiler, so any use of C99 and later constructs has to be optional (e.g. OPUS_INLINE). This is also why we do not use C++-style // comments.
To reduce the risk of exploitable memory errors, we do not use any function pointers in the code unless they are declared as static const. We also have "flat" objects, which can be copied using a "shallow copy", so do not add pointers to non-static data in the data structures.