Files
dress-pr/en-US/PARTIAL_CLONE.md
xhgitandClaude Code cf17c9ea42 Initial commit: 小南娘 - 面向可爱的蓝孩子的 git 学习实践项目
仿 Dress 的照片投稿仓库, 适配自建 Gitea 平台 git.xhmcl.com:
- README / GUIDE (PR 教程) / CONTRIBUTING / Q&A / EXIF / PARTIAL_CLONE (中英双语)
- A-Z + # 首字母目录结构
- .gitea: PR 模板、维护者名单、7 个纯 bash + API 的 Actions 工作流 (可选功能)

Co-Authored-By: Claude Code <noreply@anthropic.com>
2026-11-07 17:17:27 +08:00

4.2 KiB

Partial Clone Guide

简体中文

The XiaoNanNiang repository contains a large number of images. Using only --depth 1 skips old commits but still downloads all images in the current version. If you only intend to modify your own directory, it's recommended to combine partial clone with sparse checkout: Git first downloads commits and the directory structure, and only fetches files on demand when the selected directory is checked out.

Requires Git 2.25 or higher. Run git --version to check. The platform only serves HTTP(S) — use the https:// URLs below.

First Clone

Fork this repository on the platform first, then clone your own fork. Replace the username and directory with actual values:

git clone --filter=blob:none --sparse --depth 1 https://git.xhmcl.com/<your-username>/dress-pr.git
cd dress-pr

# Check out only your own directory; you can set it even if it does not exist remotely yet
git sparse-checkout set '<first-letter>/<your-folder-name>'

For example, to use Y/Yueosa:

git sparse-checkout set 'Y/Yueosa'

Now the working tree contains the root documents and Y/Yueosa, without other contributors' images. If the folder name starts with a digit or symbol, use the # directory and always quote the path:

git sparse-checkout set '#/233'

Next you can create directories, add photos, and use git add, git commit and git push as usual. Continue to the Beginner Guide for the full contribution workflow.

Managing Checked-Out Directories

View the currently selected directories:

git sparse-checkout list

Add another directory while keeping existing ones:

git sparse-checkout add 'A/AnotherName'

When re-selecting directories, set replaces the previous directory list:

git sparse-checkout set 'F/FirstName' 'S/SecondName'

Git downloads files in newly selected directories as needed, so the smaller the directory, the less download and disk usage.

Getting More History

--depth 1 creates a shallow clone containing only the latest commit. Normal photo submission and PRs are unaffected; if you need the full history, run:

git fetch --unshallow

This fetches the complete commit history, while image contents are still fetched on demand by the partial clone.

Restoring the Full Repository

If you just want all files of the current version in the working tree, disable sparse checkout:

git sparse-checkout disable

This downloads files not yet fetched for the current version and can be large. The repository keeps its partial clone settings, and old-version files continue to be fetched on demand.

If you really need to convert the repository to a full clone with all history and contents, continue with:

git fetch --unshallow --no-filter
git config --unset remote.origin.partialclonefilter
git config --unset remote.origin.promisor

If you already ran git fetch --unshallow, it will tell you the repository is not shallow; use git fetch --no-filter instead.

FAQ

Why do root documents still appear?

git clone --sparse uses cone mode by default. This mode keeps the repository root files so you can read README.md, GUIDE.md and other documentation directly.

Why can't I see the directory I just selected?

Run git sparse-checkout list first to check the path, case and first-letter directory. New directories that don't exist remotely won't appear automatically; they show up in the working tree only after you create and add files.

Does partial clone affect commits or PRs?

No. Files that are not checked out are still managed by Git and won't be deleted when you git add or commit your own directory. Only modify your own directory, and check your changes with git status before committing.

Push fails with an authentication error?

The platform has no SSH. Push with the https:// URL and authenticate with "username + access token". See Step 2 Method B of the Beginner Guide for how to generate a token.

References