Contributing to carpentry lessons with GitHub25 Jul 2016
Set up your remote upstream and merge updates from there
- Fork the repo you want to work on.
- Clone that repo down, e.g.,
git clone https://github.com/USERNAME/library-shell/
- So you can fetch changes from the originating repo, add remote reference there:
git remote add upstream https://github.com/data-lessons/library-shell/(you can see your remotes by
git remote -v)
- Get changes from
git fetch upstream
- Merge those changes locally:
git merge upstream/gh-pages
- Repeat Number 4 & 5 above before you begin a new unit of work below to insure you have the latest base version of the lesson
Do work in a local branch and submit changes to the lesson repo
- Start with the latest version of the upstream lesson (see above)
- Create a branch for your improvements:
git checkout -b new-lesson-improvement
- Install Jekyll if you want to preview your changes locally
make serveto preview locally, typically at http://127.0.0.1:4000/
- Once done with your work as you see fit, run
git commit, and then
git push origin new-lesson-improvementup to your forked repository.
- Make a pull request from your repo in GitHub (this tells the upstream maintainers: hey, pull my improvement into the upstream repo)
Getting set up to improve lessons
After we taught Library Carpentry here at UCSD, we sat down and worked through the workflow for contributing to Carpentry lessons. Matt Critchlow, our IT Dev Manager, walked us through the CONTRIBUTING.md document and I worked up the steps below from our meeting.
One of the confusing aspects on translating the common fork/pull-request development workflows is that most of the documentation found on the web is spelled out for
master branches. This is because, by convention, the default branch when you initialize a repository in GitHub (or locally) is named
master. However, with the Software/Data/Library Carpentry lessons the default branch is set to
gh-pages. This is mainly for ease, because the web version of the lessons live in this branch and this is where Software Carpentry wants the work to go for contributions. Also, on GitHub, commits to this branch will be processed by Jekyll, a static site generator, making the nice lesson webpages we use in class. The main thing to know is that in Software Carpentry lesson land, when you see
master in Git help or online documentation, you can mentally substitute it with
gh-pages. Hopefully, this will help folks new to git contribute more to the lessons.
Update 2016-08-28: Corrected the Getting changes from the upstream default branch
gh-pages section to remove using
git status to check changes in the upstream remote as this won’t work!
git fetch upstream is the right command to pull down any changes that may have been made.
Setup your fork and local clone
Clone the library-shell project to your local machine (USERNAME - your GitHub user account name). Having a local copy allows us to edit locally using our favorite tool, create branches for discrete work and keep the local repository in synch with
$ git clone https://github.com/USERNAME/library-shell/
clonegrabs the repository and makes a local copy. It will create the directory (named for the repository name by default) and sets up the linkages between your clone and the remote repository (called
origin). Let’s confirm this by running
git remote -v.
$ cd library-shell $ git remote -v
origin https://github.com/ucsdlib/library-shell.git (fetch) origin https://github.com/ucsdlib/library-shell.git (push)
Now configure the upstream remote. We are doing this because we want to be able to periodically grab – especially before we start our lesson improvements – new changes from the upstream repository’s (
data-lessons/library-shell) gh-pages branch and merge those with our local gh-pages branch. To set an upstream remote let’s us do this:
$ cd library-shell $ git remote add upstream https://github.com/data-lessons/library-shell/
Now, take a look at the remotes again and confirm you have an upstream fetch and push pair in addition to your
$ git remote -v origin https://github.com/ucsdlib/library-shell.git (fetch) origin https://github.com/ucsdlib/library-shell.git (push) upstream https://github.com/data-lessons/library-shell.git (fetch) upstream https://github.com/data-lessons/library-shell.git (push)
Getting changes from the upstream default branch
Before you begin your work on contributing to a lesson, you should always fetch and merge changes from the upstream repository. Why? Think about how many contributors are involved in working on some of the Carpentry lessons (hundreds). Imagine this: after you forked and cloned the
data-lessons/library-shell repository, you take a break and go out for a coffee and sandwich. While you were away a contributor made a pull request and the lesson maintainer merged it into the lesson. Now you return from your break, make your lesson changes, commit, push and then find in GitHub your work conflicts with the upstream repository. Argh! You don’t want to deal with that after a nice coffee and sandwich. So, if you begin work without incorporating the latest approved version of the lesson there will be a greater chance for conflicts when you are ready to make the pull request. Further, maybe the change made while you were away fixes the same issue you had with the lesson. To avoid working on an outdated base document, always run
get fetch and update you local repository before you start work.
Let’s fetch the changes down. We do this by fetching changes from our upstream repository.
$ git fetch upstream remote: Counting objects: 11, done. remote: Compressing objects: 100% (11/11), done. remote: Total 11 (delta 2), reused 0 (delta 0), pack-reused 0 Unpacking objects: 100% (11/11), done. From https://github.com/data-lessons/library-shell 67d2478..db5039b gh-pages -> upstream/gh-pages
Merge those changes into your gh-pages branch. Note, you are currently in your
gh-pagesbranch and the following command will merge the fetched changes from
$ git merge upstream/gh-pages Updating 67d2478..db5039b Fast-forward _episodes/01-intro-shell.md | 22 +++++++++++----------- _includes/navbar.html | 1 + contribute.md | 10 ++++++++++ 3 files changed, 22 insertions(+), 11 deletions(-) create mode 100644 contribute.md
git statusshould look like the below. Yay! We are now ready to work on the lesson.
$ git status On branch gh-pages Your branch is up-to-date with 'upstream/gh-pages'. nothing to commit, working directory clean
Contributing to the lessons
Your are ready to make improvements to the lessons! To do this, we need to create a branch in which to add our changes. Think of branches as a unit of work that hangs together. So, if you are going to add an exercise in the loop episode in the shell lesson, this should hang together as a named
branch, such as
Create a branch for changes to the lesson:
$ git checkout -b new-lesson-improvement
gitwill create the
new-lesson-improvementbranch and switch you into it.
Now, develop (work) on
new-lesson-improvementand make the improvement or addition to the lesson. How do I know it looks ok? I mean this is webcontent and I want to see what it looks like. To do this we need to have Jekyll – the static site generator that GitHub uses to prepare and render our lessons – installed locally. Follow these instructions to install Jekyll. (You can safely skip step 4 “R Packages” most likely).
Now, we can kick the tires and see what our work looks like locally, by going back to our terminal and running:
$ make serve
If it runs successfully, the output will contain a line that tells you where your server is running:
Server address: http://127.0.0.1:4000/
Navigate to that url in your browser and you should see the lesson you are working on and your changes. You can continue to work on your changes and refresh to see the changes in the browser.
But wait, why didn’t I use
jekyll serveinstead? That’s what the Jekyll documentation says. You could run
jekyll serveand it would do the same thing. The template developers for Software Carpentry have created a make file that supports a number of command in addition to running a wrapper for
jekyll serve. Go ahead and
ctr-cto kill your server and type
makeby itself on the command line and you’ll see this:
The make command contains a number of helpful commands for working with your lesson including
lesson-checkto validate markdown and
clean-rmdto clean intermedicate R files.
Once done with your work as you see fit, run
git commit, and then
git push origin new-lesson-improvementup to your forked repository:
$ git add . $ git commit -m "adding loop challenge" $ git push origin new-lesson-improvement
- Now, you can perform the Pull Request on GitHub from your account’s forked repository of library-shell. Follow these steps.
- The maintainer of the lesson will review the changes (be patient!) and merge them into the upstream repository. They might also have questions for you and in this case, will comment on your request before the changes can be merged.
- Begin beautiful contributor-maintainer iteration here.
- Matt Critchlow @mcritchlow led us through the development workflow. Juliane Schneider @pitviper6, as a maintainer for the Open Refine lesson, also was a great resource for figuring out the carpentry workflow.
- Development workflow with Git: Fork, Branching, Commits, and Pull Request
- Configuring a remote for a fork
- Syncing a Fork