Documentation/glossary.txt: describe remotes/ tracking and packed-refs
[git/git.git] / Documentation / repository-layout.txt
1 git repository layout
2 =====================
3
4 You may find these things in your git repository (`.git`
5 directory for a repository associated with your working tree, or
6 `'project'.git` directory for a public 'bare' repository).
7
8 objects::
9 Object store associated with this repository. Usually
10 an object store is self sufficient (i.e. all the objects
11 that are referred to by an object found in it are also
12 found in it), but there are couple of ways to violate
13 it.
14 +
15 . You could populate the repository by running a commit walker
16 without `-a` option. Depending on which options are given, you
17 could have only commit objects without associated blobs and
18 trees this way, for example. A repository with this kind of
19 incomplete object store is not suitable to be published to the
20 outside world but sometimes useful for private repository.
21 . You also could have an incomplete but locally usable repository
22 by cloning shallowly. See gitlink:git-clone[1].
23 . You can be using `objects/info/alternates` mechanism, or
24 `$GIT_ALTERNATE_OBJECT_DIRECTORIES` mechanism to 'borrow'
25 objects from other object stores. A repository with this kind
26 of incomplete object store is not suitable to be published for
27 use with dumb transports but otherwise is OK as long as
28 `objects/info/alternates` points at the right object stores
29 it borrows from.
30
31 objects/[0-9a-f][0-9a-f]::
32 Traditionally, each object is stored in its own file.
33 They are split into 256 subdirectories using the first
34 two letters from its object name to keep the number of
35 directory entries `objects` directory itself needs to
36 hold. Objects found here are often called 'unpacked'
37 (or 'loose') objects.
38
39 objects/pack::
40 Packs (files that store many object in compressed form,
41 along with index files to allow them to be randomly
42 accessed) are found in this directory.
43
44 objects/info::
45 Additional information about the object store is
46 recorded in this directory.
47
48 objects/info/packs::
49 This file is to help dumb transports discover what packs
50 are available in this object store. Whenever a pack is
51 added or removed, `git update-server-info` should be run
52 to keep this file up-to-date if the repository is
53 published for dumb transports. `git repack` does this
54 by default.
55
56 objects/info/alternates::
57 This file records paths to alternate object stores that
58 this object store borrows objects from, one pathname per
59 line. Note that not only native Git tools use it locally,
60 but the HTTP fetcher also tries to use it remotely; this
61 will usually work if you have relative paths (relative
62 to the object database, not to the repository!) in your
63 alternates file, but it will not work if you use absolute
64 paths unless the absolute path in filesystem and web URL
65 is the same. See also 'objects/info/http-alternates'.
66
67 objects/info/http-alternates::
68 This file records URLs to alternate object stores that
69 this object store borrows objects from, to be used when
70 the repository is fetched over HTTP.
71
72 refs::
73 References are stored in subdirectories of this
74 directory. The `git prune` command knows to keep
75 objects reachable from refs found in this directory and
76 its subdirectories.
77
78 refs/heads/`name`::
79 records tip-of-the-tree commit objects of branch `name`
80
81 refs/tags/`name`::
82 records any object name (not necessarily a commit
83 object, or a tag object that points at a commit object).
84
85 refs/remotes/`name`::
86 records tip-of-the-tree commit objects of branches copied
87 from a remote repository.
88
89 packed-refs::
90 records the same information as refs/heads/, refs/tags/,
91 and friends record in a more efficient way. See
92 gitlink:git-pack-refs[1].
93
94 HEAD::
95 A symref (see glossary) to the `refs/heads/` namespace
96 describing the currently active branch. It does not mean
97 much if the repository is not associated with any working tree
98 (i.e. a 'bare' repository), but a valid git repository
99 *must* have the HEAD file; some porcelains may use it to
100 guess the designated "default" branch of the repository
101 (usually 'master'). It is legal if the named branch
102 'name' does not (yet) exist. In some legacy setups, it is
103 a symbolic link instead of a symref that points at the current
104 branch.
105 +
106 HEAD can also record a specific commit directly, instead of
107 being a symref to point at the current branch. Such a state
108 is often called 'detached HEAD', and almost all commands work
109 identically as normal. See gitlink:git-checkout[1] for
110 details.
111
112 branches::
113 A slightly deprecated way to store shorthands to be used
114 to specify URL to `git fetch`, `git pull` and `git push`
115 commands is to store a file in `branches/'name'` and
116 give 'name' to these commands in place of 'repository'
117 argument.
118
119 hooks::
120 Hooks are customization scripts used by various git
121 commands. A handful of sample hooks are installed when
122 `git init` is run, but all of them are disabled by
123 default. To enable, they need to be made executable.
124 Read link:hooks.html[hooks] for more details about
125 each hook.
126
127 index::
128 The current index file for the repository. It is
129 usually not found in a bare repository.
130
131 info::
132 Additional information about the repository is recorded
133 in this directory.
134
135 info/refs::
136 This file is to help dumb transports to discover what
137 refs are available in this repository. Whenever you
138 create/delete a new branch or a new tag, `git
139 update-server-info` should be run to keep this file
140 up-to-date if the repository is published for dumb
141 transports. The `git-receive-pack` command, which is
142 run on a remote repository when you `git push` into it,
143 runs `hooks/update` hook to help you achieve this.
144
145 info/grafts::
146 This file records fake commit ancestry information, to
147 pretend the set of parents a commit has is different
148 from how the commit was actually created. One record
149 per line describes a commit and its fake parents by
150 listing their 40-byte hexadecimal object names separated
151 by a space and terminated by a newline.
152
153 info/exclude::
154 This file, by convention among Porcelains, stores the
155 exclude pattern list. `.gitignore` is the per-directory
156 ignore file. `git status`, `git add`, `git rm` and `git
157 clean` look at it but the core git commands do not look
158 at it. See also: gitlink:git-ls-files[1] `--exclude-from`
159 and `--exclude-per-directory`.
160
161 remotes::
162 Stores shorthands to be used to give URL and default
163 refnames to interact with remote repository to `git
164 fetch`, `git pull` and `git push` commands.
165
166 logs::
167 Records of changes made to refs are stored in this
168 directory. See the documentation on git-update-ref
169 for more information.
170
171 logs/refs/heads/`name`::
172 Records all changes made to the branch tip named `name`.
173
174 logs/refs/tags/`name`::
175 Records all changes made to the tag named `name`.
176
177 shallow::
178 This is similar to `info/grafts` but is internally used
179 and maintained by shallow clone mechanism. See `--depth`
180 option to gitlink:git-clone[1] and gitlink:git-fetch[1].
181