[Users] How to submit patches to the Einstein Toolkit
schnetter at cct.lsu.edu
Mon Nov 10 10:46:57 CST 2014
Let me follow up on my explanation, which may have seemed lengthy to
some. The upshot is:
- Almost all of the steps I describe you need to do only once, such as
e.g. create a Bitbucket account, or fork the repository.
- "Pull requests" work in the following way: Instead of giving
everybody write access to the same repository, people have their own
repositories. That's what git is all about! Instead of pushing to the
main repo, they push to their own repo, and then create a pull
request. This is only one button push more complicated than pushing to
the main repository.
On Thu, Nov 6, 2014 at 11:47 AM, Erik Schnetter <schnetter at cct.lsu.edu> wrote:
> When using Subversion, there are two classes of people: Those with
> write access to the repository, which can make changes or create
> branches for development, and those without, who are forced to send
> patches or tarballs by email or by attaching them to a bug report. The
> latter is really quite inconvenient, both for the people preparing
> them, and also for the people who have to apply them.
> When using a git repository hosted at Bitbucket (or Github etc.),
> things are much easier for all parties. There is no need to send
> around patches or tarballs any more. Instead, you do the following.
> Bear with me -- the instructions are lengthy only because I'm
> explaining things, but each of these steps really corresponds just to
> a mouse click or two.
> (1) Create an account at Bitbucket. Obviously, you do this only once.
> (2) Fork the repository on Bitbucket. This creates a copy of the
> repository on Bitbucket which belongs to you, and where you can commit
> changes. I see the "fork" menu entry hidden behind the three
> horizontal dots in the top-left corner of the original repository's
> web pages.
> That's "fork", and not "clone" -- clone would make a copy on your
> laptop, which isn't quite enough. The difference is that you can later
> allow others to look at your forked repository, and thus pick up the
> changes you made.
> (3) Clone the forked repository, or -- if you already have a clone of
> the original repository -- add a new "remote" to it. That's
> straightforward if you use a GUI, on the command line you'll have to
> follow instructions that are posted on Google.
> All the steps until here are necessary only once per repository.
> (Unfortunately, the ET uses a gazillion of repositories -- in
> hindsight I'm not quite sure why, it certainly doesn't make things
> (4) Make your changes, commit them, etc. I usually create a new branch
> for every patch.
> (5) Push your changes (i.e. the branch) to the forked repository. They
> are now visible to the world; people can look at them and comment
> (6) Tell others about your changes and where to find them. The easiest
> way is to create a "pull request" on Bitbucket. I see the "create pull
> request" menu entry near the "fork" menu entry. A "pull request" is
> essentially a ticket (on Bitbucket) that allows others -- those with
> write access to the repository -- to merge ("pull") your changes with
> the click of a button. Those with write access like this, because it's
> very convenient. They don't even have to clone your branch -- it's all
> visible right there in the web browser, and reviewing and accepting a
> patch takes less than a minute.
> As an added bonus, since you created the commit, your name will be
> associated with the change and bring you fame.
> And as a side note, my (egoistic) willingness to look at a patch
> depends on how much trouble I have to go through for this. Clicking on
> a link that immediately shows me a nice commit message in a web
> browser together with a colour-coded diff is about as easy as it gets
> for me.
> Erik Schnetter <schnetter at cct.lsu.edu>
Erik Schnetter <schnetter at cct.lsu.edu>
More information about the Users