Please, before sending patches, read these brief comments. They are here to help the project have both its users happy using the program and the developers/maintainers feel good when trying to change code that other people contributed.
For the record, when this document mentions "I", it mostly means Rogério Theodoro de Brito's (@rbrito) is the one to blame.
When you write your pull request and your commit messages, please, be detailed, explaining why you are doing what you are doing. Don't be afraid of being too verbose here. Also, please follow the highly recommended guidelines on how to write good good commit messages.
When in doubt, follow the model of the Linux kernel commit logs. Their commit messages are some of the best that I have seen. Also, the ffmpeg has some good messages that I believe that should be followed. If you are in a hurry, read the section named "Contributing" from subsurface's README.
Make sure that you have all dependencies installed, like via:
pip install -r requirements.txt
pip install -r requirements-dev.txt
Run the test suite with
py.test coursera/test -v --cov coursera --cov-report html
If some test fails, please don't send your changes yet. Fix what broke before sending your pull request.
If you need to change the test suite, explain in the commit message why it needs to be changed (e.g., the page layout or the authentication methods from coursera changed, or they implemented a new kind of course).
Running only py.test is okay for one-time check, but as it runs on your currently active Python installation, you're only checking the script on one Python version. A combination of tox + pyenv will allow you to run tests on many Python versions easily.
First, install tox (listed in requirements-dev.txt) and pyenv (see their manual: pyenv installation. Make sure to activate pyenv or start a new shell. Then, install Python versions and run tox:
pyenv install 3.5.1
pyenv install 3.4.4
pyenv install 3.3.6
pyenv install 3.2
pyenv install 2.7.5
pyenv install 2.6.9
pyenv local 3.5.1 3.4.4 3.3.6 3.2 2.7.5 2.6.9
tox
Please, help keep the code tidy by checking for any potential bugs with the
help of flake8, and pylint. If you know of any
other good tools for analyzing the code, let me know about them!
If you happen to find any issue reported by these programs, I welcome you to fix them. Many of the issues are usually very easy to fix and they are a great way to start contributing to this (and other projects in general). Furthermore, we all benefit from a better code base.
If you are proposing the use of a substitute of a tool that we already use, take a few paragraphs to tell us why we would like to change.
If we are not using something, it is most likely that one of the following options applies:
- I (@rbrito) may not even know that what you are proposing exists or have not yet "seen the light" as to why I should use it instead of using what I am currently using.
- Even if I know about the tool, I may not know how to use it, or how it would make me more productive. Educate me and we all will gain from a better project.
Fork the repository using github (there's a button) and clone your copy to
your local machine for your personal modifications. Let's say that your
github username is username:
git clone https://github.com/username/coursera-dl
Initiate a new branch on your local repository copy.
cd coursera-dl
git checkout -b my-new-branch
Then, start hacking, doing whatever you want. Really, just play with what you want. After you make each "logical chunk" of changes (say, you fixed one specific bug, or you made one specific feature), commit your changes.
Use the following commands to see which files you changed and which changes you made, just to recap, before sending them to the public (up to this moment, everything that you made lives only on your computer):
git status
git diff
You may have made more changes than you want to contribute at this point in time (this is very common). So, we select which changes we want to commit with:
git add -p
Then a little prompt will appear and you will interactively select the
changes that you want to make part of the commit. The main options there are
y or n for including/not including respectively and s to split a very
large change into smaller changes ones (this is a killer feature of git add -p).
To see the changes that you have selected to be part of your commit, use:
git diff --staged
See if everything is OK. If not, then use git add -p to add something that
you may have missed or, if you included something else that you didn't
intend to include, "un-include" it with:
git reset -- file-with-changes-to-un-include
Then, you can start over and include only what you meant to put in a single commit.
If you think that your changes make one "logically" cohesive change to the project, create a commit of them in the repository with:
git commit
Then, enter a descriptive message of what (not why, not how) you changed on the first line and after that, you are free to write your prose detailing how you did what you did (e.g., which documentation you read), why you did what you did (Is that a preparation for a future change? Does it improve the speed or memory requirements of the project?) and other comments.
Once again, it is good to have in mind some rules for good commit messages.
To see the commit that you just made (with author, date, short and long description only), use:
git show
If you want to see the entire commit, use:
git show -p
If you are satisfied with your changes, go ahead and type:
git push origin my-new-branch
Where my-new-branch is the branch that you created for this set of changes
(you remember that, right?). If everything goes correctly, then, after a few
seconds, your changes will be right there on the cloud, on github! Yay!
To inform the maintainer of the project that you have some shiny changes that you are proposing for inclusion on the main project, github will show you somewhere on your fork of the project that you have a branch and that you can create a pull request.
Click that button, compose a message for the maintainer of the project and tell him/her about what you just did, why you are proposing etc. Send that message and the maintainer will be notified that you made changes and that you'd like them to be included in the main project.
That's essentially it.
There are many other tricks/steps in the git workflow, but these are the basics that I (@rbrito) think that will suffice for a start. If you want a few details more, feel free to ask me to include them here.
This section is for project maintainers.
DRAFT
- Run tests locally. Use tox to run tests on all supported Python versions.
- Run the script on several courses to check sanity
- Update CHANGELOG.md, increment version, put what you've added to the
changelog into commit message. This way it gets it way into releases page
with a nice description of the changes.
git add ... & git ci -m 'Bump version (old_version -> new_version)' git tag new_versiongit push && git push --tagspython setup.py sdist bdist_wheel --universalto build the packagetwine upload dist/coursera-dl-0.6.1.tar.gzto deploy the package.
Build new Docker image from PyPI package:
docker build --tag courseradl/courseradl --build-arg VERSION=0.11.2 .
Run the image:
docker run --rm -it -v "$(pwd):/courses" -v "$HOME/.netrc:/netrc" courseradl -n /netrc -- google-machine-learning
Publish the image:
docker push courseradl/courseradl