1GIT repository layout 2===================== 3v0.99.5, Sep 2005 4 5You may find these things in your git repository (`.git` 6directory for a repository associated with your working tree, or 7`'project'.git` directory for a public 'naked' repository). 8 9objects:: 10 Object store associated with this repository. Usually 11 an object store is self sufficient (i.e. all the objects 12 that are referred to by an object found in it are also 13 found in it), but there are couple of ways to violate 14 it. 15+ 16. You could populate the repository by running a commit walker 17without `-a` option. Depending on which options are given, you 18could have only commit objects without associated blobs and 19trees this way, for example. A repository with this kind of 20incomplete object store is not suitable to be published to the 21outside world but sometimes useful for private repository. 22. You can be using `objects/info/alternates` mechanism, or 23`$GIT_ALTERNATE_OBJECT_DIRECTORIES` mechanism to 'borrow' 24objects from other object stores. A repository with this kind 25of incompete object store is not suitable to be published for 26use with dumb transports but otherwise is OK as long as 27`objects/info/alternates` points at the right object stores 28it borrows from. 29 30objects/[0-9a-f][0-9a-f]:: 31 Traditionally, each object is stored in its own file. 32 They are split into 256 subdirectories using the first 33 two letters from its object name to keep the number of 34 directory entries `objects` directory itself needs to 35 hold. Objects found here are often called 'unpacked' 36 objects. 37 38objects/pack:: 39 Packs (files that store many object in compressed form, 40 along with index files to allow them to be randomly 41 accessed) are found in this directory. 42 43objects/info:: 44 Additional information about the object store is 45 recorded in this directory. 46 47objects/info/packs:: 48 This file is to help dumb transports discover what packs 49 are available in this object store. Whenever a pack is 50 added or removed, `git update-server-info` should be run 51 to keep this file up-to-date if the repository is 52 published for dumb transports. `git repack` does this 53 by default. 54 55objects/info/alternates:: 56 This file records absolute filesystem paths of alternate 57 object stores that this object store borrows objects 58 from, one pathname per line. 59 60refs:: 61 References are stored in subdirectories of this 62 directory. The `git prune` command knows to keep 63 objects reachable from refs found in this directory and 64 its subdirectories. 65 66refs/heads/`name`:: 67 records tip-of-the-tree commit objects of branch `name` 68 69refs/tags/`name`:: 70 records any object name (not necessarily a commit 71 object, or a tag object that points at a commit object). 72 73HEAD:: 74 A symlink of the form `refs/heads/'name'` to point at 75 the current branch, if exists. It does not mean much if 76 the repository is not associated with any working tree 77 (i.e. 'naked' repository), but a valid git repository 78 *must* have such a symlink here. It is legal if the 79 named branch 'name' does not (yet) exist. 80 81branches:: 82 A slightly deprecated way to store shorthands to be used 83 to specify URL to `git fetch`, `git pull` and `git push` 84 commands is to store a file in `branches/'name'` and 85 give 'name' to these commands in place of 'repository' 86 argument. 87 88hooks:: 89 Hooks are customization scripts used by various git 90 commands. A handful of sample hooks are installed when 91 `git init-db` is run, but all of them are disabled by 92 default. To enable, they need to be made executable. 93 94index:: 95 The current index file for the repository. It is 96 usually not found in a naked repository. 97 98info:: 99 Additional information about the repository is recorded 100 in this directory. 101 102info/refs:: 103 This file is to help dumb transports to discover what 104 refs are available in this repository. Whenever you 105 create/delete a new branch or a new tag, `git 106 update-server-info` should be run to keep this file 107 up-to-date if the repository is published for dumb 108 transports. The `git-receive-pack` command, which is 109 run on a remote repository when you `git push` into it, 110 runs `hooks/update` hook to help you achive this. 111 112info/grafts:: 113 This file records fake commit ancestry information, to 114 pretend the set of parents a commit has is different 115 from how the commit was actually created. One record 116 per line describes a commit and its fake parents by 117 listing their 40-byte hexadecimal object names separated 118 by a space and terminated by a newline. 119 120info/rev-cache:: 121 No higher-level tool currently takes advantage of this 122 file, but it is generated when `git update-server-info` 123 is run. It records the commit ancestry information of 124 the commits in this repository in a concise binary 125 format, and can be read with `git-show-rev-cache`. 126 127info/exclude:: 128 This file, by convention among Porcelains, stores the 129 exclude pattern list. `git status` looks at it, but 130 otherwise it is not looked at by any of the core GIT 131 commands. 132 133remotes:: 134 Stoers shorthands to be used to give URL and default 135 refnames to interact with remote repository to `git 136 fetch`, `git pull` and `git push` commands.