Git Version Control Kevin Scannell David Ferry CSCI

Git & Version Control Kevin Scannell, David Ferry CSCI 5030 – Principles of Software Development Saint Louis University St. Louis, MO 63103 1

Version Control 101 • Project files (source code, etc. ) are stored in a “repository” and are shared with all developers. • All changes are “committed” to the repository with a descriptive log message. • Developers work independently, but commit meaningful increments back to the project when they are finished. CSCI 5030 – Principles of Software Development 2

Historic Reasons for Version Control • Shared, up-to-date code base – There is no “master” copy of the code – Multiple programmers can work on the same code simultaneously – Committed code is always backed up • Tracks and attributes all changes to code – All changes have a log message that records why the change was made • These are things that programmers did before version control, just manually CSCI 5030 – Principles of Software Development 3

Modern Reasons for Version Control Modern techniques have evolved around the use of version control. • Allows branching of project – e. g. stable branch, experimental branch, specific release branches • Feature branches – can develop each feature in it’s own branch with “pristine” code base. When feature is done and tested it is merged • Allows “diff” between any two versions of code • Allows debugging and “root cause analysis. ” Can find the exact commit (and why) that introduced a bug • Allows “blame view” – can see files with each line annotated by the person who added it and the log message • Bigger systems (like Git. Lab) incorporate project planning, code reviews, etc. into repository CSCI 5030 – Principles of Software Development 4

Scope of Version Control Meant for source code – Works really well for plain text files Also: data files, config files, build scripts, test code, documentation, READMEs, license, the project web site, etc. – Again, works great when things are plain text Generated products are not stored – E. g. object files and other intermediate build results, including final binary Other project’s source code and libraries are up for debate – Replicating where code lives is usually not great – What if your build script automatically fetches remote libraries if they’re not present on your system? – Dependencies and version numbers should at least be documented CSCI 5030 – Principles of Software Development 5

Centralized VC Systems First VC systems were centralized “clientserver” systems. • Only one official master copy • All clients communicate to-from that official master • The official master has control: e. g. can enforce policy by “locking” files, etc. These are systems such as CVS and SVN CSCI 5030 – Principles of Software Development 6

Distributed VC Systems Distributed VC adopts a peer-to-peer model • Every copy of the repository contains the entire repository and its entire history • No master copy- all copies are equal – Don’t confuse “master branch” in Git with a “master repository” • Developers can work locally using VC features without having to talk to any upstream VC server – Multiple local copies can all have revisions, must be manually synchronized (e. g. with git pull) • Allows for different workflows that don’t make sense in centralized VC – Integrator workflow, dictator & lieutenants, etc. CSCI 5030 – Principles of Software Development 7

Git • Linus Torvalds in 2005 (where else have I heard that name…? ) • Needed a fast, distributed, and safe version system for the Linux kernel • Stack. Overflow survey in 2018 said 88% of professional developers used Git in their work (next highest was SVN at 17%) CSCI 5030 – Principles of Software Development 8

Git in a Picture (e. g. at SLU) git. cs. slu. edu Ex: on machine 1 git pull origin master <make changes> git commit * -m “xyz” git push origin master Remote (origin) Machine 1 Working Copy Machine 2 Local Working Copy Local CSCI 5030 – Principles of Software Development 9

Git in a Picture (e. g. at SLU) git. cs. slu. edu Ex: on machine 1 git pull origin master <make changes> git commit * -m “xyz” git push origin master Remote (origin) Machine 1 Working Copy Machine 2 Local Working Copy Local CSCI 5030 – Principles of Software Development 10

Basic Git Terminology • origin – the remote repository that you originally cloned the project from • master – the master branch • remotes – copies of the repository that are stored on a remote server (contrast with local copies that are stored on each developer’s machine) • working copy – the file system where developers look at and modify code on their own machine (but is not the local repository) • branches – “copies” of the master branch used for developing new features • forks – real copies of the repository used for developing new features CSCI 5030 – Principles of Software Development 11

Common Git Commands • git clone – fetch a new repository and create a new working copy of that repository • git add – add files to version control • git commit – save changes to the local repository • git push – send changes from the local repository to a remote • git fetch – update your local repository without modifying the working copy • git pull – update your local repository and merge those changes with the working copy CSCI 5030 – Principles of Software Development 12