NotesGitGit

Git

Git

back to main index

General

  • git is based on stroring snapshots, not differences

  • every operation is local

  • everything is checksumed (SHA-1)

  • a lifecycle of every file is following: File lifecycle

  • index is proposed next commit snapshot, that means all files that will go in next commit

  • typical workflow looks following: Workflow

  • files matches by pattern in .gitignore are ignored

  • --cached == --staged

  • master for default branch and origin for default remote are just naming conventions, there is nothing special about those names

  • commits as a pointer to snapshot which can consist of many trees (one tree represent one directory), trees are holding pointers to blobs which represent files Commit

  • each commit points to a snapshot and parent commit Commits

  • branch is a pointer to specific commit, and nothing more than that

  • HEAD is last commit snapshots, normally it points to some branch

  • Branch

  • detached head means that HEAD is not pointing to any branch

  • all files(blobs) are stored as object under SHA-1 hash

  • if files are common between snapshots, they are shared: Snapshot

Workflow

  • to initialize repo use git init
  • to check status of files use git status
    • git status -s for short output
  • git add adds files to the index, it adds the file to staging area, but it is also used to start tracking files (as side effect kinda)
  • to see how file how been modified use git diff
    • to see what diffs are stages use git diff --staged
  • to commit what is staged in index, use git commit -m "commit message"
    • to stage all modified files in one step add -a flag: git commit -a -m "message"
  • to remove file use git rm file, it does 2 things:
    • untracks file
    • removes file from the harddrive
  • to untrack file but keep it on disk use git rm --cached <file>
  • to move file use git mv <fromFile> <toFile>
  • to see the commit history use git log
    • git log --graph prints pretty tree
  • to modify the last commit that has not been pushed yet, use git commit --amend
    • it can update the files and commit message
  • to unstage staged file on unmodify modified file use git restore <file>,
    • basic version unmodifies file while git restore --staged <file> unstages
  • to work with remotes there is git remote command, in basic form it displays added remote
    • for move verbose output use git remote -v
    • to add remote git remote add <name> <url>
    • to fetch info from given remote git fetch <name>, you can also pull by git pull <name>
    • to get more detailed information about given remote, like branch tracking info use git remote show <remote>
    • renaming: git rename <oldName> <newName>
    • removing: git remote remove <remote>
  • tags are pointers to specific commits, there are 2 types of tags in git:
    • lightweight which are similiar to branches, there are just pointers
    • annotated tags which are objects and can store additional data
  • refer to docs as there are a lot of commands
  • keep in mind that tags must be pushed separately, they are not pushed when branch is pushed
  • git aliases are a way to create shortcut(alias) for existing command
  • to check what is active branch use git branch
    • for get more info use git branch -all
    • to get all tracking information use git branch -vv
  • to create new branch use git branch <name>
  • to switch between branches (repoint HEAD) use git checkout <name>
  • to create a branch and checkout to it in one step use git checkout -b <name>
    • to create a new branch based on one on remote use git checkout -b <name> <remote/name>
      • the shortcut for that is git checkout --track <remote/name>
      • or even git checkout <name> if there is one remote with given branch name
  • to merge different branch to current branch use git merge <name>
  • to delete give branch use git branch -d <name>
  • to rename branch use git branch --move <oldName> <newName>
  • to push specific branch use git push <remote> <branch>
    • to set tracking on remote add --set-upstream flag
  • rebasing is applying series of commits to another branch since common parent commit
    • to rebase go to branch that you want to rebase from and use git rebase <branchRebased>
      • you dont need to checkout to target branch by git rebase <targetBranch> <branchRebased>
      • usually after rebase you will want to fast-forward merge the target branch to move the pointer git merge <targetBranch>
    • rebase can be viewed as alternative to merge
    • you can also rebase if there is no common parent commit: git rebase --onto <targetBranch> <branchWithCommonCommit> <branchBeingRebased>
  • while git checkout resets to which branch HEAD is pointing, git reset resets to which commit the branch pointed by HEAD is pointing
    • great explanation is in proGit book
    • git reset --soft resets only to which commit the branch is pointing, index and working directory are left untouched
    • git reset [--mixed] which is default option, does the previous stuff + it resets the index to snapshot which HEAD is now pointing
    • git reset --hard does the previous + resets also working directory to new index, which may LOSE the data, hence its the only one reset that may be dangerous
Built with LogoFlowershow