Version Control with Git

Talk on Git from January 2011 CoMO Rich Web meet up

Transcript

1. Version Control With Git A tale forked merged and bludgeoned by Clint Shryock
2. A brief history
3. • Changes passed around via patches and archive files • Started using BitKeeper in 2002 • Things were good mostly... • Allegation of license violations • Free license revoked in 2005 • Git project was born (hurray!) Linux Kernel Maintenance A Brief History
4. • Speed • Simple design • Strong support for non-linear development (thousands of parallel branches) • Fully distributed • Able to handle large projects like the Linux kernel efficiently (speed and data size) Design Goals A Brief History
5. How Git Works
6. How Git Works Thinking About Data: SVN Model • It’s a series of tubes! diffs • Operations usually include network overhead • Offline support: edit but can’t commit
7. How Git Works Thinking About Data: SVN Model • It’s a series of tubes! diffs • Operations usually include network overhead • Offline support: edit but can’t commit • A Sad Keanu
8. How Git Works Thinking About Data: Git Model • A set of snapshots • Generally doesn’t require any network connection • Can edit commit branch merge rewind
9. How Git Works Thinking About Data: Git Model • A set of snapshots • Generally doesn’t require any network connection • Can edit commit branch rewind • Happy Keanu!
10. How Git Works Thinking About Data: Git Model • 3 states: unmodified, modified, staged • 3 sections: Git directory, working directory, staging area
11. Branching
12. Branching • SHA-1 checksum • Pointer to the snapshot • Author and message metadata • Zero or more pointers to direct parents First: What’s in a commit?
13. Branching When you commit Example: Stage 3 files • Checksum for each file • Store that file version in the Git repository • Adds checksum to staging area Commit those files • Git checksums each subdirectory • Stores those tree objects in the Git repository • Create a commit object with metadata and pointer to the root project tree (you liked it so you put a SHA-1 on it)
14. Branching Commit disected
15. Branching A Series of Commits
16. Branching What’s a branch already
17. Branching What’s a branch already
18. Branching Special pointer: HEAD
19. Branching Special pointer: HEAD $ git branch testing
20. Branching Special pointer: HEAD $ git checkout testing
21. Branching Special pointer: HEAD $ vim test.rb $ git commit -a -m “A change”
22. Branching Special pointer: HEAD $ git checkout master
23. Branching Special pointer: HEAD $ vim test.rb $ git commit -a -m “Another change”
24. Demos!
25. Demos! • add • modify • stage • commit • branch • merge Demos • log • log ..master log master.. • ignore • diff (--cached) • stash • merge from remote • branch from remote's fetch
26. Collaboration
27. • Remote repositories are hosted versions of your project • Possible to have several remotes • Some push some pull Remotes Collaboration
28. • Clone • Fetch • Pull • Rebase Getting data Collaboration
29. • References to the state of branches on your remote repositories • Moved automatically whenever you fetch • origin/master origin/feature/awesome-feature Remote Branches Collaboration
30. Working with Remote Branches Collaboration
31. Working with Remote Branches Collaboration $ git clone clint@git.ourcompany.com:project.git
32. Working with Remote Branches Collaboration
33. Working with Remote Branches Collaboration
34. Working with Remote Branches Collaboration $ git fetch origin
35. Working with Remote Branches Collaboration $ git merge origin/database
36. Git on the Server
37. • Remote repositories are typically bare • Several protocols to choose from • local /opt/git/project.git • git:// • ssh:// • http(s):// Hosting Git Repositories on the Server Git on the Server
38. • Repository is in another directory on disk • Works well if your team has access to shared filesystem • direct patch vs. file:// • Pros • leverage existing file permissions and network access • can pull directly from each other, bypassing remotes • Cons • limited by network access • NFS mount can be slower than SSH on the same server Local Protocol Git on the Server
39. • Pros • SSH is easy to setup • All data transfer is encrypted and authenticated • Efficient: data as compacted before transfer • Cons • Can’t serve anonymous access SSH Protocol Git on the Server
40. • Pros • Fastest transfer protocol available • Uses same data-transfer mechanism as SSH protocol • Can serve anonymous access • Cons • Lack of authentication • More difficult to setup Git Protocol Git on the Server
41. • Pros • Simplicity • Uses same data-transfer mechanism as SSH protocol • Can serve anonymous access • Cons • Inefficient for the client • No* write access HTTP(s) Protocol Git on the Server
42. Version Control With Git Images that weren’t total hacks were used with consent courtesy of Scott Chacon http://scottchacon.com/ @ctshryock ctshryock.com github.com/ctshryock/GitTalk github.com/progit progit.org twitter www sources