Notes for Authors

Getting and working with the sources

Get an account at Github

This step is not absolutely necessary if you have other means to make your private development visible to the public.

  1. Register a new github account. Let’s assume your account name is foofoo.
  2. Sign in to your github foofoo account.
  3. Go to the official SymbolicData repository and click on Fork. You will now have your own clone of the SymbolicData repository.

How to get the sources

Every developer should have a public and a private repository. (See section about integration manager workflow below.)

You should clone from your public repository.

git clone

How get future updates from the official repository

You have to connect your repository with the official one. Add a new remote with name upstream. Do this inside your cloned repository.

git remote add upstream git://

Now, by using

git fetch upstream

you get all the commits from the official repository.

Note that after fetch you only have the commits locally on your computer. Whether and how you merge them, you can decide after looking at the current situation. Usually

gitk --all

gives you a nice graphical picture of the DAG (directed acyclic graph) of the commits.

Workflow: how to work with the repositories

The following assumes that the SymbolicData project is working with the Integration Manager Workflow.

There is a blessed repository and each developer has a public repository (for example, at github as described above) and a private repository on his laptop/desktop computer.

Code would flow via PR (pull requests) to the blessed repository. Github has a button for pull requests.

Advantage of this workflow is that the integration manager can act as a person to ensure the quality of the patches. And there is no need to manage ssh keys in order to give someone write access to the repository.

If this sounds too complicated for a small project, we can also switch to a centralized workflow. That should be discussed on the mailing list.

Make yourself known to git

If in the output of git config --list you don’t find something like LastName

, then you should introduce yourself to git. This will create/modify a file ~/.gitconfig.

git config --global "YourFirstName YourLastName"
git config --global ""

You should use your default email address. This email address is only used by git to identify you, but not to send any mail to you.

Structure of the repository and how to deal with it

The SymbolicData project decided to follow “A successful Git branching model”.

The blessed repository consists of 3 branches.

master This branch
is for all the data that should go into version 3 of SymbolicData and marks the latest stable release.
This branch is for actual development. Main requirement for a commit on this branch is that programs must compile and data must be in a consistent state (no syntactic errors).

This branch is basically a snapshot from the latest Mercurial repository (12-Dec-2012). It contains the full history of former SymbolicData commits.

This branch will eventually be removed.

In fact, there are two disconnected DAGs for master, and old-master. (Something that cannot be done with Mercurial.)

How to work with GIT

It is usually enough to work with the following git commands:

git pull
Get the latest changes from the official repository.
git add FILE2.m
Mark and FILE2.m to be committed in the next commit.
git commit

Commit all marked modifications to the local repository.

Note that only files that have been added via git add will be committed.

Note that after git commit the modifications have not yet left your computer. They are simply stored in your local repository.

Set the variable GIT_EDITOR in your $HOME/.bashrc if you don’t like vi so much.

export GIT_EDITOR="emacs -nw"
git push
Upload your modifications to the official repository.
git status
Investigate which files have changed on your file system.
git log
Investigate the history of changes.