docs(blogs): add git usage cheatsheet

- Cover git update-index --skip-worktree for ignoring local
  changes to tracked files
- Include stashing, amending, undo, reflog, bisect, log, clean
  and cherry-pick tips
- Add matching zh-hans and en-gb versions
This commit is contained in:
zihluwang committed 2026-08-20 15:09:25 +08:00
1 parent 25a50486cb
commit 7f6f3d7a00
2 files changed
+322

No files matched your search

+161
View File
@@ -0,0 +1,161 @@
---
title: Git Cheatsheet
tags:
- git
- tips
author:
name: Zihlu Wang
email: real@zihluwang.me
---
## Ignoring Local Modifications to Tracked Files: `git update-index --skip-worktree`
When you need to modify a Git-tracked file locally but don't want those changes to show up in `git status` or in commits (e.g. a local `application-local.yml`, `.env`, or IDE configuration), you can use `--skip-worktree` to make Git ignore local modifications to that file.
### Marking and Unmarking
```bash
# Make Git ignore local modifications to the file
git update-index --skip-worktree <file>
# Unmark the file and track it again
git update-index --no-skip-worktree <file>
```
### Listing Ignored Files
Marked files are shown with an `S` flag in `git ls-files -v`:
```bash
git ls-files -v | grep '^S'
```
### Committing Changes to a Marked File
To commit new changes to the file, unmark it first, commit, then mark it again:
```bash
git update-index --no-skip-worktree <file>
git add <file>
git commit -m "chore: update config"
git update-index --skip-worktree <file>
```
### `--skip-worktree` vs `--assume-unchanged`
Both options stop Git from reporting modifications to tracked files, but they serve different purposes:
| Option | Purpose |
| -------------------- | ----------------------------------------------------------------------------------------------------------------- |
| `--assume-unchanged` | A performance optimisation for large files that rarely change; Git may re-track the file once it notices a change |
| `--skip-worktree` | Deliberately ignore local changes for "keep local modifications, don't commit" scenarios; more persistent |
So for "ignoring local changes", use `--skip-worktree`.
### Caveats
- `--skip-worktree` only affects the working tree — it is not a permanent ignore. Operations that rewrite the file, such as `git pull`, `git checkout`, or `git stash`, may overwrite your local changes or cause a conflict. If the file is also likely to change upstream, unmark it before pulling.
- It only applies to **tracked** files; for untracked files, use `.gitignore`.
## Stashing Unfinished Work: `git stash`
When you need to switch branches mid-task, use `git stash` to set your working-tree changes aside:
```bash
git stash push -m "work in progress" # save with a message
git stash -u # include untracked files
git stash list # list saved stashes
git stash pop # restore and drop the stash
git stash apply # restore but keep the stash
git stash drop # delete a stash
```
## Amending the Latest Commit: `git commit --amend`
When you commit and then realise you forgot a file or want to change the message:
```bash
git add <forgotten-file>
git commit --amend # keep the same message, fold in new changes
git commit --amend -m "new message" # also change the message
```
Note: if the commit has already been pushed, prefer adding a new commit instead of rewriting shared history.
## Undoing Changes: `git restore` and `git reset`
```bash
git restore <file> # discard unstaged working-tree changes
git restore --staged <file> # unstage but keep the changes (like git reset HEAD <file>)
git restore --source=HEAD <file> # restore the file to its HEAD version
```
To discard all uncommitted changes (including the staging area):
```bash
git reset --hard HEAD
```
`--hard` also resets the working tree, so double-check that nothing you want to keep would be lost.
## Recovering Lost Commits: `git reflog`
When a commit seems to have vanished after a bad `git reset --hard` or `git rebase`, `reflog` records every place HEAD has been:
```bash
git reflog
git reset --hard HEAD@{2} # go back to an earlier recorded position
```
As long as the commit still exists in the object database, reflog can get you back to it.
## Bisecting a Regression: `git bisect`
When you don't know which commit introduced a bug, `git bisect` uses binary search to locate it:
```bash
git bisect start
git bisect bad # the current commit is bad
git bisect good <sha> # mark a known-good commit
```
Git checks out a midpoint commit; you just mark each as `good` or `bad` until it narrows down the offending commit:
```bash
git bisect good # or git bisect bad
git bisect reset # finish and return to your branch
```
## Friendlier History: `git log`
```bash
git log --oneline --graph --decorate --all # graph view across all branches
git log -S "some-string" --all # find commits that added/removed a string
git blame <file> # show the last modifier of each line
git blame -L 10,20 <file> # restrict to a line range
```
## Cleaning Up Untracked Files: `git clean`
```bash
git clean -n -fd # preview which files would be removed (always run this first)
git clean -fd # delete untracked files and directories
```
`-n` only shows what would be deleted without deleting it — preview before you execute.
## Picking Commits: `git cherry-pick`
Apply a specific commit from another branch onto your current branch:
```bash
git cherry-pick <sha>
```
## Deleting Merged Local Branches
```bash
git branch --merged | grep -v '\*' | xargs -n 1 git branch -d
```
This deletes all local branches merged into the current branch (except the current one). Preview with `git branch --merged` first.