NEURON contribution guidelines
Table of contents
There are several separate repositories that make up the NEURON project and you may contribute to any of these. A single code board manages the full set. This Contributing document is relevant not only for this repository but also for these other repositories.
Five major repositories are directly included at this github:
- nrn -- the main NEURON code. When you fork and clone this repository you will find the actual source in subdirectories under nrn/src. The subdirectories of primary interest will generally be those most recently edited -- nrniv, nrnmpi, nrnpython. By contrast some of the subdirectories will rarely if ever need updating.
- progref-py -- NEURON Python programmer's reference. As mentioned further below, the beginning user can often be of particular help to the community by clarifying documentation and examples which may have seemed clear to the person who wrote it but are lacking details that are needed for a student.
- tutorials -- here again, the student or beginning user can be of assistance either by improving existing tutorials or by providing new tutorials to illuminate aspects of NEURON that he or she had trouble with.
- iv -- the old Interviews graphical program, now largely deprecated in favor of modern GUIs.
- progref-hoc -- this is the programmers reference for the hoc language, the 'higher order calculator' of Kernighan (a co-founder of UNIX). hoc is now largely retired in favor of Python. The hoc code itself is in the nrn repository and also should not usually be touched.
There are several NEURON-related repositories hosted elsewhere which also encourage contributions. Each of these will have it's own Contributing document.
- NMODL -- improved method for compiling and porting .mod files using abstract syntax trees
- CoreNEURON -- an optimized NEURON for running on high performance computers (HPCs)
- NetPyNE -- multiscale modeling tool for developing, simulating and analyzing networks which include complex cells, potentially with detailed molecular-modeling.
- NetPyNE-UI -- graphical user interface for NetPyNE
We encourage contributions to the NEURON simulator from individuals at all levels -- students, postdocs, academics, industry coders, etc. Knowledge of the domain of neural simulation is also helpful but much of the simulation technology is comparable to other simulation fields in biology and beyond -- numerical integration of ordinary differential equations (ODEs), here coupled with events (event-driven).
If you want to pick up and try an existing improvement project, you will note that we have indicated levels of difficulty with labels. Most internal hacks will require knowledge of C/C++. Knowledge of python is also necessary for writing accompanying test code. Note that it can be much easier to get started by improving documentation or by adding new tutorials.
There are some things that will make it easier to get your pull requests accepted quickly into the master build where it is teed-up for eventual release.
Before you submit an issue, search the issue tracker to see if the problem or something very similar has already been addressed. The discussion there might show you some workaround or identify the status of a project.
Consider what kind of change you have in mind:
For a Major Feature, first open an issue and make clear whether you are request the feature from the community or offering to provide the feature yourself. The code board will consider whether this is needed or whether some existing feature does approximately the same thing. If you would like to yourself implement the new feature, please submit an issue with the label proposal for your work first. This will also allow us to better coordinate effort and prevent duplication of work. We will also then be able to help you as you develop the feature so that it is can be readily integrated.
Small Features can be submitted directly.
When you're ready to contribute to the code base, please consider the following guidelines:
- Make a fork of this repository. From there you will clone your own version of the repository to your machine for compilation and running.
- Make your changes in your version of the repo in a new git branch:
git checkout -b my-fix-branch master
- For Python code contributions follow the Python Contribution guidelines.
- When creating your fix or addition, consider what new tests may be needed for continuous integration (CI), and what new documentation or documentation change is needed.
- The full test-suite will run once you submit a pull-request. If you have concerns about the code, or if you've already failed the test-suite, you can run the test-suite locally on your machine.
- Pull (fetch and merge) from upstream so that your current version contains all changes that have been made since you originally cloned or last pulled.
git pull --ff upstream master
- Commit your changes using a descriptive commit message.
git commit -a
- Push your branch to GitHub; note that this will push back to your version of NEURON code in your own github repository (the origin).
git push origin my-fix-branch
- In GitHub, send a Pull Request to the
masterbranch of the upstream repository of the relevant component -- this will run the test-suite before alerting the team.
- If we suggest changes then:
- Make the required updates.
- Rewrite; Rerun test-suites; Repush; Rereq
After your pull request is merged, you can safely delete your branch and pull the changes from the main (upstream) repository.
Development conventions: NEURON code is being built according to C/C++, Python best-practices. The easiest way to figure out what this is is to take a look at current code and copy the way things are formatted, indented, documented, and commented.
The Neuron Development Topics section of the documentation provides a starting point for understanding NEURON's internals.
- Make sure to install clang-format and cmake-format with Python version you are using:
pip3.7 install cmake-format==0.6.0 pyyaml --user brew install clang-format # or your favorite package manager
Now you should have the
cmake-format commands available.
-DNRN_CMAKE_FORMAT=ONoption of CMake to enable CMake code formatting targets:
cmake .. -DPYTHON_EXECUTABLE=`which python3.7` -DNRN_CMAKE_FORMAT=ON
With this, new target called cmake-format can be used to automatically format all CMake files:
$ make cmake-format Scanning dependencies of target cmake-format Built target cmake-format
You can now use
git diff to see how cmake-format has formatted existing CMake files.
Note that if you want to exclude specific code section to be not formatted (e.g. comment blocks), you can use guards:
# ~~~ # This comment is fenced # and will not be formatted # ~~~
# cmake-format: off # This bunny should remain untouched: # . ＿ ∩ # cmake-format: on
See cmake-format documentation for details.
clang-formatyou should restrict formatting to only the code parts relevant to your change. This can be eachieved by setting following build options:
cmake .. -DNRN_CLANG_FORMAT=ON \ -DNRN_CMAKE_FORMAT=ON \ -DNRN_FORMATTING_ON="since-ref:master" \ -DNRN_FORMATTING_CPP_CHANGES_ONLY=ON
Note: Sometimes it might be necessary to point your build-system to the clang-format-diff utility,
this can be done by supplying an additional flag:
- You can then run the
clang-formattarget after a full build:
make && make clang-format
The Python source code is located under
share/lib/python/neuron. Python unit tests are
test/pynrn and use the pytest framework, along with pytest-cov for code coverage. Code coverage report can be found on codecov.io under share/lib
pre-commit hooks, the
black tool can be executed every time you commit, and automatically reformat your code according to the repository's preferences.
To set up automatic formatting run the following commands from the root of the repository:
pip install pre-commit pre-commit install
Note: The first time you commit after installing the hooks, it can take a while to set up the environment. If your commit contains any formatting errors, an error will be displayed: Black will have reformatted your staged changes and your workspace will contain them as new unstaged changes; stage these changes and re-commit the now formatted code!
You can also install and manually apply black formatting as follows:
pip install black black /path/to/file.py
This will reformat the entire file, not just your changes, and might pollute your PR.
Have you tested on the current alpha version of NEURON? If not, please clone, compile, install and run with current version (this one).
Please let us know what operating system you were using when you found the bug. If you have access to another operating system, it is helpful if you can find out if the bug shows up there as well. Please indicate which operating system(s) the bug has been found in.
In order to address a bug quickly, we need to reproduce and confirm it. Usually bugs will arise in the context of a simulation which will in many cases be extremely large. Please provide a simple, short example (zipped with any associated mod files and a README). The simplifying process may require considerable work to isolate the bug arising somewhere in the midst of a large simulation. Sometimes, this process alone is enough to identify the bug as a function limitation or documentation insufficiency, rather than a code bug per se.
Pull requests are reviewed by the development team. If you don't receive any feedback after a couple weeks, please follow up with a new comment.