This document introduces basic workflows for managing code assignments with Git and CodeTub in the CS 251 Fall 2019 computing environment.

To start the tutorial, skip to Concepts.

Contents

Quick Reference

To start the tutorial, skip to Concepts. This section is for reference once you know the basics.

Solo Workflow

To start the assignment:

  1. cs251 start assignment --solo to create your hosted submission repository and clone it locally.

As you work:

  1. git add and git commit cohesive changes often
  2. git push commits often

To submit your work:

  1. git add and git commit your final changes
  2. cs251 sign your respository to sign your work and respond to any assignment survey questions.
  3. git push all commits to your hosted submission repository. git status should show both:
    • Your branch is up to date with 'origin/master'.
    • nothing to commit

Team Workflow

To start the assignment:

  1. cs251 start assignment and enter the team’s usernames to create (or connect) your team’s hosted submission repository and clone it locally.

As you work:

  1. Coordinate with teammates before making changes to avoid conflicts.
  2. git add and git commit cohesive changes often
  3. git pull changes by your teammates and:
  4. git push changes often

To submit your work:

  1. git add and git commit your final changes
  2. cs251 sign your respository to sign your work and respond to any assignment survey questions.
  3. git push all commits to your hosted submission repository. git status should show both:
    • Your branch is up to date with 'origin/master'.
    • nothing to commit

Concepts

Git is a version control system. Version control systems are designed to help structure and automate the otherwise tedious task of recording the way a project changes through several distinct versions over time, created by one or more users contributing separate changes to the project.

CodeTub is a system for managing code assignments with version control.

Git

A Git repository is a structured history of a project’s files through multiple explicitly recorded snapshots or versions.1 Each snapshot is called a commit (noun) or a revision. Each commit is accompanied by metadata describing which earlier commits it follows, who created it and when, plus a description of the important changes since the last commit.

A history of committed versions of a repository.
A project's version history as commits to a repository.

A local Git repository is paired with a working copy2 of the project files in a directory where the user can prepare and commit changes.3 The git tool lets users inspect the repository history as well as the status of the working copy. Each time the user edits the working copy to create a meaningful new version of the project, they run git commands to record new commits in the repository.

A local repository and its working copy.
A local repository and its working copy.

Repositories can be cloned to create a new repository with the same history. New commits can be created in separate clones, then shared by pushing and pulling commit history between related repository clones to bring them up to date with each other.

Pushing and pulling commits between repository clones.
Pushing and pulling commits between repository clones.

git command-line tool

Type along.

To learn your way around these tools, you need to follow along and execute commands in this tutorial yourself. This tutorial provides a sample assignment repository that you will use to run your own commands. Fire up a terminal and type along!

Operations on Git repositories are accomplished by running commands of the form git command in the working copy. Run git with no command to get a summary of available commands.

Take note of what it shows (omitted here) and the fact that you can summon this at any time. The command git help topic summons help about a topic such as a command. Try git help help to get help with git help!

CodeTub

CS 251 uses CodeTub to distribute, collect, and grade assignments using Git repositories.

For each assignment, the CodeTub repository host stores a private submission repository for each team (or individual).4 This repository is where you will save and submit (i.e., commit and push) intermediate and final versions of your work. When you are done, the instructors will collect the latest version of your work directly from this hosted submission repository for grading. It is accessible only by you, your teammates, and the instructors.

To create and work with hosted submission repositories, you will use the command-line CodeTub client tool (cs251) and the command-line git tool as described in the rest of this document.

First Steps

This tutorial assumes that Wendy (username wendy) is logged in and will work with Wanda (username wanda) as a pair. Work with a partner and replace these two throughout the tutorial with your own names and usernames.

We use the example assignment/lab repository tutorial to demonstrate necessary tasks working with repositories in CS 251. Please type along through this tutorial to try all of the steps yourself.

Authenticate your account

$ is the command prompt

Command line transcripts shown in this document use $ to represent the command prompt. Do not type it.

Do these steps once per user per computing environment in your own account.

For now, Wendy (representing the user who is logged in) should do these steps.

(If Wanda can log in separately and complete the account setup now, great, but later is fine too. Wanda must do this when logged in as Wanda, not as Wendy!)

  1. If you have not done so already, run the general setup and installation steps for the computing environment you are using.

  2. Once your account is configured to use the CS 251 environment, authenticate your account with CodeTub for access to hosted repositories by running the command:

    $ cs251 auth -i f19

    This command will prompt for your human name, CS username, and password. It may take a moment to complete.

List hosted repositories

To see a list of your repositories stored remotely on the CS 251 CodeTub repository host, run the command:

$ cs251 list

Even though you have not created any submission repositories yet, you should see output like the following, showing that you have Read access to the courses/cs251/f19/starters/tutorial repository, which holds starter code for the tutorial exercises, including this tutorial.

wendy's repositories hosted by git@cs.wellesley.edu:

 R  	courses/cs251/f19/starters/tutorial

As you create submission repositories, they will also appear in this list.

Now it’s time to start your first assignment and clone your first repository.

Start an assignment (create a submission repository): cs251 start

Optional Extra: git clone

Under the hood, cs251 start uses the git clone command to clone the entire history of your hosted CodeTub repository to the local machine and check out a working copy of the latest commit.

You do not need to use git clone directly in CS 251, since start does it for you. If you are curious, after you have created your hosted repository with cs251 start, you can try it out directly: git clone the repository using the URL listed by cs251 start.

$ cd cs251-repos
$ git clone ssh://git@cs.wellesley.edu/courses/cs251/f19/submissions/wanda-wendy/tutorial another-tutorial

Note the extra argument another-tutorial which gives an explicit name for the local clone.

Let’s assume you have directory called cs251-repos where you want to keep all your CS 251 work. If you do not yet have such a directory, create one (just once) like this:

$ mkdir cs251-repos

Do this once per team per assignment.

First cd into your directory for CS 251 assignments.

$ cd cs251-repos

To start work on an assignment or lab activity named, e.g., tutorial, use the command:

$ cs251 start tutorial

This command does a few things:

  1. Prompts for the usernames of the teammates (if any) with whom you will work on this assignment or lab activity.
    • For this tutorial, you should be working in pairs, so enter your partner’s username when prompted.
    • If you have no teammates, just hit enter without typing anything at the teammate prompt.
  2. Creates a hosted submission repository for your team on the CS 251 CodeTub repository host.
  3. Initializes this repository with the relevant starter code.
  4. Clones your hosted repository to your local computer and creates a working copy in a directory, tutorial in this case, where you can make changes.

After running the start command, take a look around the working copy directory that was created and inspect the starter code.

$ ls
tutorial
$ ls tutorial
...

If you ever accidentally delete your local working copy or just forget that you already started an assignment, it is OK to run cs251 start again (with the same teammates). It will reuse the hosted repository you created earlier.

Solo Basics

This section introduces the basic CodeTub and Git tasks necessary to work with CS 251 assignment repositories for individual assignments by stepping through a sample assignment. The Team Basics section covers CodeTub and Git tasks for collaboration on team assignments as well as other more advanced issues.

Inspect history: git log

You should now have a clone of the tutorial repository along with a working copy on your local machine. The next few parts will use only on your local copy of the repository. We will not interact with your hosted repository again until pushing changes.

Change directory into the working copy base directory and list its contents.

$ cd tutorial
$ ls
define.rkt README.md

Remember, these files are a working copy of the latest commit in the repository, but they tell you nothing about the history.

Run the command git log to obtain a summary of the revision history. (Yes, actually run it!) The output is a log of commit records. (You might see something slightly different in your repository.)

$ git log
commit cc2bb091418a2b3333905303ac06bee5b0e92de5 (HEAD -> master, origin/master)
Author: wendy <wendy@tutorial.wellesley.edu>
Date:   Mon Sep 17 10:23:09 2018 -0400

    wendy started team wanda-wendy

commit d6ef986bb3e1b360aeffb23c3892dee969a7f9f2 (starter/master)
Author: Dorothy Vaughn <dvaughn@acd.nasa.gov>
Date:   Fri Aug 31 15:08:26 1962 -0400

    Add survey.

commit 303fab4b27829954fee87979fee4df654a78ee23
Author: Dorothy Vaughn <dvaughn@acd.nasa.gov>
Date:   Fri Aug 31 15:07:19 1962 -0400

    Add a .gitignore file.

commit 382f18c3b561594cd3437b6e4940ed15be6def99
Author: Dorothy Vaughn <dvaughn@acd.nasa.gov>
Date:   Fri Aug 31 14:40:10 1962 -0400

    Add explanations in the 

commit 45359e0c1deb1f3c66ce930f6bba06e3c0d376f9
Author: Dorothy Vaughn <dvaughn@acd.nasa.gov>
Date:   Fri Aug 31 14:39:04 1962 -0400

    Starter code for Git and CodeTub Tutorial

Paged Output

If the output of git log or other git commands is longer than the terminal window, it will be shown using a pager (less). You can tell by whether you get another command prompt or if the output ends with a colon on its own line. If the output is paged, you can use the space bar to scroll forwards a page at a time (and b to scroll back), just like with less. When you are done, type q to quit the pager and return to the command prompt.

Each commit is identified by a large hexadecimal number that permanently identifies this commit in this repository’s history.

The latest commit was made automatically to record your team membership when you ran cs251 start tutorial. The earlier commits were made by course staff in preparing the starter code repository. Notice that two commit are labeled with some references in parentheses.

  • The reference HEAD indicates the “most recent” commit on which the working copy is based.
  • The reference master refers to the main branch of development. In CS 251, we will use only this branch.
  • The reference origin/master describes state of your hosted submission repository; starter/master shows the last of the commits you got from the starter repository. For now, we will ignore these references.

History often happens to be a linear (i.e., totally ordered) sequence of commits, but is a directed acyclic graph (i.e., partially ordered) in general. Each commit (except for the first) is descended from one or more earlier commits, called its parents, representing the preceding version(s) of the repository. To see the topological order of commits more clearly, run git log with the --graph option:

$ git log --graph

The working copy is the intuitive next commit in the making, not yet recorded in history, so it does not show up in the log.

Inspect changes: git status, git diff

While editing the working copy, it is useful to inspect its differences with the most recent revision in the repository.

Summarize changes: git status

Ignore most (hints) from git status.

The parenthesized hints in the output of git commands like git status or a failing git commit include a mix of useful and less useful options. CS 251 students are probably best served by ignoring most of git’s hints for now.

git status compares the working copy to its parents to determine what files have been changed in the working copy. Try it!

$ git status
On branch master
Your branch is up to date with 'origin/master'.

nothing to commit, working tree clean

Indeed Git sees we have made no changes.

Let’s make a change. Add your name in a comment at the top of define.rkt and save it. Now check again to see if Git has noticed your changes.

$ git status
On branch master
Your branch is up to date with 'origin/master'.

Changes not staged for commit:
  (use "git add <file>..." to update what will be committed)
  (use "git checkout -- <file>..." to discard changes in working directory)

    modified:   define.rkt

no changes added to commit (use "git add" and/or "git commit -a")

Git gives us a lot of information here, including several hints about actions we could take. The key information is on these two lines:

Changes not staged for commit:
    ...
    modified:   define.rkt

These indicate that the file define.rkt has been modified since the last commit but has not (yet) been marked for inclusion in the next commit. Ignore the parenthesized hints in this output for now.

Let’s explore more before returning to interpret the other lines.

Examine detailed changes: git diff HEAD

git diff HEAD shows the exact difference between the current working copy and the last commit, also known as HEAD. Lines prefixed by - are deletions; lines prefixed by + are additions. Did Git detect your change? As with git log, the output of git diff may be paged.

Prepare and record a commit: git add and git commit

The process of committing (recording) new changes to the repository history involves two steps:

  1. Select (or stage) the changes in the working copy to record with git add; and
  2. Create a new commit from those selected (staged) changes with git commit.

Let’s try it out. Run git add define.rkt to select all of your changes to define.rkt for inclusion in your pending commit.

The Commit Rule: commit cohesive changes.

One commit should record one logical, independent, working change – an atomic unit of work. As with code style, this sense can take time and practice to develop. Good rules of thumb:

  • Keep unrelated changes in separate commits; keep small mutually dependent changes in a single commit.
  • Avoid committing code that does not compile, especially if working with others. If you need to commit partial or broken changes, consider commenting out the incomplete pieces.
  • Commit often enough that you will not lose much work if your working copy suddenly disappears and you need to recover a previously committed version.
$ git add hello.c

No output means success. To confirm that Git has staged our changes hello.c in preparation for commit, run git status.

$ git status
On branch master
Your branch is up to date with 'origin/master'.

Changes to be committed:
  (use "git reset HEAD <file>..." to unstage)

    modified:   define.rkt

Notice that, compared to the earlier git status output, something has changed. Specifically, where git status previously reported define.rkt in the section:

Changes not staged for commit:

It now shows define.rkt under a section:

Changes to be committed.

Excellent! We can now use git commit to record this change to define.rkt in the local repository history. (It does not affect any other repositories.) We use the --message or shorthand -m flag to specify a commit message describing these changes. (If we omit the message on the command line, git will open an editor5 to record the commit message describing the commit.)

$ git commit -m "Added name to define.rkt"
[master af51490] Added name to define.rkt
 1 file changed, 2 insertions(+), 1 deletion(-)

The output summarizes the commit that was recorded into the repository history.

Review: git add and git commit

To record changes, do both of these steps together:

  1. git add all files with changes you wish to commit.
  2. git commit to record these changes in a commit.

Shortcut: git commit -a -m "..." or git commit -am "..."

You can often replace the two-step git add-git commit dance by the single command git commit -a. The --all or -a flags tell git commit to automatically git add all the changes to existing files and then commit. If you wish to commit a new file for the first time, use git add.

In case you forget to git add

A common mistake when learning Git is to forget to git add changes in preparation for a commit. In this case, git will give a reminder message that you have no changes added to commit. The message also acknowledges that you have made changes in the working copy, but that they were not staged for commit by git add:

$ git commit -m "Added name to define.rkt"
On branch master
Your branch is up to date with 'origin/master'.

Changes not staged for commit:
	modified:   define.rkt

no changes added to commit

If this happens, just go ahead and git add the changes and git commit again.

Push changes: git push

So far, we have cloned your hosted submission repository to create a local repository and committed some changes to the local repository. We have not yet modified the hosted submission repository.

Now that we have committed a change, let’s check git status again.

$ git status
On branch master
Your branch is ahead of 'origin/master' by 1 commit.
  (use "git push" to publish your local commits)

nothing to commit, working tree clean

The Push Rule: push often.

Push your new local commits to the remote repository reasonably often (at least every time you leave the computer), even if only as a backup. (What if your computer hits a meteorite while you are out?)

When sharing a repository, pushing often allows your teammates to stay more up to date on your work.

Notice that the status indicates that there is nothing to commit. It also indicates that Your branch is ahead of 'origin/master' by 1 commit, meaning that there is one commit in our local repository that is not in the remote (origin) repository. Indeed, we just made that commit!

git push copies local commits to a remote repository. By default, git pushes to the same remote repository from which the local repository was cloned (in this, case your hosted submission repository, which is exactly what we want).

git push your changes to your hosted repository now. When working with others, pushing may require more work first, but it should succeed without any coordination when working alone on one account (as in this tutorial so far).

$ git push
Enumerating objects: 5, done.
Counting objects: 100% (5/5), done.
Delta compression using up to 24 threads.
Compressing objects: 100% (3/3), done.
Writing objects: 100% (3/3), 309 bytes | 309.00 KiB/s, done.
Total 3 (delta 2), reused 0 (delta 0)
To ssh://cs.wellesley.edu/courses/cs251/f19/submissions/wanda-wendy/tutorial
   cc2bb09..e928079  master -> master

Submit work: cs240 sign and push

Submit: The course staff will collect your work directly from your hosted repository. To submit your work:

  1. Make sure you have committed your latest changes.

    $ git add ...
    $ git commit ...
  2. Run the command cs251 sign to sign your work and respond to any assignment survey questions.

    $ cs251 sign
  3. Push your signature and your latest local commits to the hosted repository.

    $ git push

Confirm: All local changes have been submitted if the output of git status shows both:

  • Your branch is up to date with 'origin/master', meaning all local commits have been pushed
  • nothing to commit, meaning all local changes have been committed

Resubmit: If you realize you need to change something later, just repeat this process.

Summary

Review the Solo Workflow section for a workflow summary.

Team Basics

This section assumes familiarity with the Solo Basics. To work with a team, each team member clones their shared CodeTub repository, then pushes and pulls revisions through it.

We need a new command to make this work: git pull acquires new revisions from a remote repository and attempts to merge them into the local repository. As discussed earlier, git push shares local revisions with a remote repository.

In this section, lab partners will now split up to play the separate roles of Wendy (the student who is logged in right now) and their partner Wanda. It is OK if Wendy and Wanda both work in the same CS account as Wendy, but you can also log in separately if another workstation is available. We assume all parties involved are working in accounts where the course setup has been done.

Wanda Make a new clone of the same CodeTub repository that Wendy cloned.

  • If you can work on a separate computer nextdoor, great!
  • If another computer is not available, it is OK if they are on the same computer and even if they are both in Wendy’s local account (though all commits will appear to come from Wendy). The two clones just need to be separate directories. For example, working in the same account as Wendy for this tutorial, Wanda could create a clone called tutorial-wanda by running:

    $ cs251 start tutorial tutorial-wanda

    Note the extra argument declaring the directory to create when cloning. Make sure to enter the same team that you did before to ensure you get the same hosted repository. To be 100% sure, type the usernames of all teammates (both wendy and wanda), even if the prompt says, e.g., “except wendy”.

Incorporate remote changes: git pull

git pull brings new revisions into the local repository from a remote repository and attempts to merge with the latest commit and update the working copy.

Fast-forward ordered changes

Wendy After Wanda has cloned the repository, commit and push a new change to the hosted repository.

Wanda Run git pull to get Wendy’s changes.

$ git pull
remote: Counting objects: 3, done.
remote: Compressing objects: 100% (2/2), done.
remote: Total 3 (delta 0), reused 0 (delta 0)
Unpacking objects: 100% (3/3), done.
From ssh://git@cs.wellesley.edu/courses/cs251/f19/submissions/wanda-wendy/tutorial
   cc8aaa3..f71eeb3  master     -> origin/master
Updating cc8aaa3..f71eeb3
Fast-forward
 define.rkt           | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

The Pull Rule: commit before pulling.

git pull attempts to change your your working copy if it finds new commits remotely. If you have uncommitted changes in your working copy, it could be confusing to get those mixed up with new remote changes. Committing your changes first means you can restore your current working copy later if needed.

Note the line of output that says Fast-forward. This is Git’s way of saying that:

  1. All of the new commits it pulled follow the current commit on which Wanda’s local working copy is based…
  2. So it went ahead and updated the repository and the working copy with the results of those new commits.

Wanda Now check for Wendy’s changes in the working copy. Also, run git log --graph to take a look at the history and confirm that the new commits appeared.

Both partners now have the save version of all files. That was easy! It out worked so easily because at the time we pulled, the history in Wanda’s repository was strictly a prefix of the history in the CodeTub repository, that had been updated most recently by Wendy. This is not always the case, as we will see.

Merge divergent changes: git pull

Wanda Edit the first line of define.rkt then git add and git commit your change. Run git pull to see if Wendy has pushed any more changes. (She has not.) git push your change to the remote repository.

Wendy Edit define.rkt to add a new line of text at the end of the file, then git add and git commit your change. Next, git pull to get any changes Wanda has pushed.

Whoa! It looks like your editor has just popped up. Before doing anything with that, flip back over to the shell and examine git’s output so far.

$ git pull
remote: Counting objects: 3, done.
remote: Compressing objects: 100% (3/3), done.
remote: Total 3 (delta 2), reused 0 (delta 0)
Unpacking objects: 100% (3/3), done.
From cs.wellesley.edu/courses/cs251/f19/submissions/wanda-wendy/tutorial
   d87c14e..c78ade8  master     -> origin/master
Auto-merging define.rkt
Merge made by the 'recursive' strategy.
 define.rkt | 1 +
 1 file changed, 1 insertion(+)

Notice the line Auto-merging define.rkt. git has figured out how merge Wendy’s and Wanda’s changes in this file automatically! Because it was successful in auto-merging all changed files, git has just popped up your editor asking for a commit message to describe the merge. Flip back to the editor, write a message, save, and then close the editor.

Take a look at define.rkt with less or emacs. Did git do the right thing? How long would it have taken you to do this manually? Step back and take a moment to think how awesome this is!

Are you back? We have a bit more to do.

Wendy Run git log --graph to see how this merge and commit affected history. We are now back to one head (revision without children) and it is the parent of Wendy’s working copy. History has become a directed acyclic graph, but we have reconverged on a single history containing both Wendy’s and Wanda’s changes.

Perspective

Consider: in a few commands, git accomplished the entire tedious process of copying Wanda’s files to Wendy’s computer, painstakingly determining which files have been updated by each person, and getting the newest revision of each file in place. In fact, it can merge changes in different files and even in different parts of the same file. Wow!

Excitement aside, automated merging is not magic nor is it a license to edit haphazardly without coordinating with your team. Even when a merge succeeds at a textual level, the end result may not be meaningful. This means it is still important to:

  • communicate with your teammates about what everyone is editing so you know what to expect from merges; and
  • do some testing or other sanity-checking on the result of a merge.

If you do not, you might still run into trouble…

Resolve conflicts

Two divergent changes to the same lines in a file constitute a conflict that cannot be merged automatically. Git is good, but it cannot read your team’s collective mind to decide what to do with a conflict, so it asks you. Try it:

Wendy Wanda Both users should now edit the first line of define.rkt in your respective local repositories to replace the comment text on that line with something more interesting. Make sure that all three of the previous version, Wendy’s version, and Wanda’s version of `` differ in the first line. Both of you should now git add and git commit your change in your respective local repositories.

Wendy git push.

Wanda git pull, then observe:

$ git pull
remote: Counting objects: 3, done.
remote: Compressing objects: 100% (2/2), done.
remote: Total 3 (delta 0), reused 0 (delta 0)
Unpacking objects: 100% (3/3), done.
From cs.wellesley.edu/courses/cs251/f19/submissions/wanda-wendy/tutorial
   f71eeb3..d87c14e  master     -> origin/master
Auto-merging define.rkt
CONFLICT (content): Merge conflict in define.rkt
Automatic merge failed; fix conflicts and then commit the result.

Wanda Note several cues in the output: Git tried, but Auto-merging define.rkt failed with CONFLICT (content): Merge conflicts in define.rkt. Finally it reports exactly what we need to do now: Automatic merge failed; fix conflicts and then commit the result.

Wanda Run git status to get some extra information.

$ git status
On branch master
You have unmerged paths.
  (fix conflicts and run "git commit")

Unmerged paths:
  (use "git add <file>..." to mark resolution)

    both modified:      define.rkt

no changes added to commit (use "git add" and/or "git commit -a")

Notice the mentions of unmerged paths.

Git tried to merge everything automatically, but Wendy and Wanda made conflicting changes to the same line(s) of a file, so there is no obvious way to merge and preserve both of their changes. Instead of doing something arbitrary, Git did the following:

  • It did not prepare define.rkt as part of a pending commit of the merge, instead instructing you to fix the conflicts and commit yourself.
  • It marked the conflicts in define.rkt using special notation to help you do a manual merge.

Let’s do that!

Wanda Opening up define.rkt, we see something like this:

<<<<<<< HEAD
/* CS 240: A simple first C program awesomified by Wanda. */
=======
/* CS 240: A simple first C program by Wendy "I seize C's seas" Wellesley. */
>>>>>>> d87c14eb957db15267cf516c1e86a72a095f0aed

If you really want to discard one person’s changes entirely:

  • git checkout --ours -- define.rkt keeps your version of define.rkt and discard your teammate’s changes to it.
  • git checkout --theirs -- define.rkt does the opposite.
  • git checkout --merge -- define.rkt provides the (un)merged version with conflicts.
  • To apply these commands to the entire working copy, omit the arguments -- define.rkt.

Just like when you edit the conflicts directly, you must continue to commit your resolved version.

The conflict markers <<<<<<< HEAD, =======, and >>>>>>> d87c14... delineate conflicting changes to the same line(s):

  • The first part, between <<<<<<< HEAD and =======, is the recent local changes to this line by Wanda. Remember HEAD refers to the latest commit in the local repository, the one on which the working copy is based.
  • The second part, between ======= and >>>>>>> d87c14..., is from the other competing changes to this line, commit d87c14... by Wendy.

Git has left it up to us to decide how to resolve these conflicting changes.

Wanda To resolve the conflict, edit the file to remove these markers and save whatever final contents you wish. It is wise to take your time and do this carefully.

Once you have resolved the conflicts to your liking, run git add define.rkt to mark define.rkt as resolved and then git commit to commit the merged version with all conflicts resolved.

$ git commit -m "Merged conflicting changes to header of define.rkt"
Recorded resolution for define.rkt.
[master e45bf94] Added name to define.rkt

The Merge Rule: isolate merge commits.

Merge commits should include only changes immediately related to merging.

  1. Commit before pulling.
  2. If pulling results in conflicts, resolve any conflicts and commit before making any other changes.

Wanda Note that we have followed the merge rule: if Git is unable to auto-merge, you should commit immediately once you have finished merging resolving any conflicts, before making any unrelated changes. The commit including a merge should include only changes immediately related to merging.

Other merge tools

Git can be connected to many other tools (including modes in Emacs and vi) that allow you to review conflicts one at a time and click or type a keyboard command to select one version or another, or type in a new one. See other resources for more.

Push changes: git push

Wanda To complete the cycle, push your changes (especially if there was just a merge) to share with your team. If there are no more new remote changes that came in since your last pull, the push should succeed. (That should be the case in this example.)

If another teammate has pushed changes to your hosted repository since you last pulled, you will need to pull and merge those changes before you can push. (Git detects that case.)

$ git push
To cs.wellesley.edu/courses/cs251/f19/submissions/wanda-wendy/tutorial
 ! [rejected]        master -> master (fetch first)
error: failed to push some refs to 'cs.wellesley.edu/courses/cs251/f19/submissions/wanda-wendy/tutorial'
hint: Updates were rejected because the remote contains work that you do
hint: not have locally. This is usually caused by another repository pushing
hint: to the same ref. You may want to first integrate the remote changes
hint: (e.g., 'git pull ...') before pushing again.
hint: See the 'Note about fast-forwards' in 'git push --help' for details.

Git gives good advice here: pull and try again. Once your local repository is up-to-date with the hosted repository, the push should succeed.

$ git push
Counting objects: 4, done.
Delta compression using up to 4 threads.
Compressing objects: 100% (3/3), done.
Writing objects: 100% (4/4), 214 bytes | 814.00 KiB/s, done.
Total 4 (delta 3), reused 0 (delta 0)
To cs.wellesley.edu/courses/cs251/f19/submissions/wanda-wendy/tutorial
   f0ae678..e45bf94  master -> master

Submit work

Use the solo submission process, but be sure to check these criteria for all teammates who have clones of the repository.

Summary

Review the Team Workflow section for a workflow summary.

Occasional Tasks

These tasks are common in general software development, but rare in CS 251 assignments.

Organize files: git add, git rm, git mv

Git repositories track revisions to files only if they are explicitly added to the repository in some commit.6

Create a new file, README, with some lucid, insightful note inside.

$ git status
On branch master
Untracked files:
  (use "git add <file>..." to include in what will be committed)

	README

nothing added to commit but untracked files present (use "git add" to track)

As git status reports, README is untracked – it is not present in the latest commit. git commit (even git commit -a) will ignore it. As the git status message helpfull indicates, we can use git add to include it in a commit, just as with changes to an existing file.

Git tracks files, not directories, so you cannot add an empty directory, but you can add a non-empty directory, which will add of the files in that directory (and recursively in any subdirectories).

The git rm command removes a file from the working copy immediately (just like the standard rm command) and also marks the file as deleted in the next commit.

git mv A renames (moves) a tracked file A to be tracked under a new name B. Like git rm, the effect is not recorded in the history until the next commit, but is reflected immediately in the working copy.7 If you move a tracked file yourself, Git notices:

Recover from mistakes

One benefit of version control is the ability to recover from mistakes (often easily) that would be time-consuming to reverse or downright disastrous in a world without version control. This section just scratches the surface of possibilities. If something goes wrong, talk to your instructor (or check documentation but act with care). It is possible to recover from a wide variety of mistakes as long as you have committed and pushed often.

Restore the committed version of a file

Overwrite the file define.rkt with something bogus.

$ echo bogus > define.rkt
$ git status
On branch master
Changes not staged for commit:
  (use "git add <file>..." to update what will be committed)
  (use "git checkout -- <file>..." to discard changes in working directory)

	modified:   define.rkt

no changes added to commit (use "git add" and/or "git commit -a")

Hey, might as well delete it.

$ rm define.rkt
$ git status
On branch master
Changes not staged for commit:
  (use "git add/rm <file>..." to update what will be committed)
  (use "git checkout -- <file>..." to discard changes in working directory)

	deleted:    define.rkt

no changes added to commit (use "git add" and/or "git commit -a")

Actually, we need that file for some more examples! Git sees that it is deleted from the working copy, but after all, we do have the entire repository history available. We can check out a fresh copy the file from the repository. We want the latest committed version (HEAD, if you recall from before), so we use the following command:

$ git checkout HEAD -- define.rkt
$ git status
On branch master
nothing to commit, working tree clean

Caution: wield git checkout with care

While we were able to recover the earlier version of define.rkt by checking it out again, git checkout itself does not make any backups of the working copy before checking out and old version of a file. That means that if we have important changes in the working copy of Makefile and we accidentally run git checkout HEAD -- Makefile, Git happily replaces our changed Makefile with the Makefile from the last commit and all those important uncommitted changes are permanently gone.

Be careful out there.

Other Resources

As you outgrow this context, consult the wealth of general Git documentation elsewhere.

License

Creative Commons License
Git for Code Assignments by Benjamin P. Wood at Wellesley College is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

This document is based on an earlier version for Mercurial with third-party hosting.

  1. Git stores history as snapshots of versions, rather than differences between versions. Each can be reconstructed from the other. 

  2. Git terminology often refers to this area for preparing new changes as the working directory, but we use the alternative term working copy to avoid confusion with the notion of working directory in the shell, as in pwd, and reinforce that it is indeed a copy. 

  3. The repository itself is hidden away in the .git directory. 

  4. Since “hosted submission repository” is a bit of a mouthful, you could also refer to this as a “hosted repository”, “CodeTub repo”, etc. 

  5. This editor is Emacs on CS 251 environments, and in general is specified by environment variable EDITOR

  6. There are many good reasons to track only explicitly added files. One is that compilers and other tools create a lot of automatically generated files that are derived purely from other files in the repository. 

  7. git mv A B is actually just mv A B; git rm A; git add B