For maintainers

Adding new indicators

The process of adding a new indicator to the library consists of several steps and is documented here for future reference.

  1. When creating a new indicator, start by declaring what type of indicator it is. For example, the rsi indicator is a momentum indicator, and hence should be implemented in the momentum.cc file. The first line of the docstring in the .cc file should be Implementation of INDICATOR, i.e. Implementation of SMA. This is needed for the linking of source code to github, since docs/source/conf.py manually retrieves the source code from github and searches for that line to get the correct line number. If the indicator have dependencies from other functions in other script, be sure to include the header files and update qufilab/indicators/CMakeLists.txt. This is needed when building manually, since the linking won’t work otherwise.

  2. When implementing the indicator it is a good idea to create the interface for it in qufilab/indicators/interface.py, where all interfaces to the indicators exist. Start by creating a non-complete interface such that one can test the actual indicator. When implementation is done, create full docstring for the indicator.

  3. When the implementation is done, the documentation for the indicator needs to be created. Start by adding the indicator to the correct entry in the docs/source/indicators.yaml file. This file ensures that correct links to the underlying source code can be made when building the docs.

  4. Add the correct specifications to the docs/source/indicators.rst directory. Observe that both an entry for the autosummary and the autofunction needs to be made and that these should be placed in alphabetic order.