1 gitrepository-layout(5)
2 =======================
6 gitrepository-layout - Git Repository Layout
15 A Git repository comes in two different flavours:
17 * a `.git` directory at the root of the working tree;
19 * a `<project>.git` directory that is a 'bare' repository
20 (i.e. without its own working tree), that is typically used for
21 exchanging histories with others by pushing into it and fetching
24 *Note*: Also you can have a plain text file `.git` at the root of
25 your working tree, containing `gitdir: <path>` to point at the real
26 directory that has the repository. This mechanism is often used for
27 a working tree of a submodule checkout, to allow you in the
28 containing superproject to `git checkout` a branch that does not
29 have the submodule. The `checkout` has to remove the entire
30 submodule working tree, without losing the submodule repository.
32 These things may exist in a Git repository.
35 Object store associated with this repository. Usually
36 an object store is self sufficient (i.e. all the objects
37 that are referred to by an object found in it are also
38 found in it), but there are a few ways to violate it.
40 . You could have an incomplete but locally usable repository
41 by creating a shallow clone. See linkgit:git-clone[1].
42 . You could be using the `objects/info/alternates` or
43 `$GIT_ALTERNATE_OBJECT_DIRECTORIES` mechanisms to 'borrow'
44 objects from other object stores. A repository with this kind
45 of incomplete object store is not suitable to be published for
46 use with dumb transports but otherwise is OK as long as
47 `objects/info/alternates` points at the object stores it
50 objects/[0-9a-f][0-9a-f]::
51 A newly created object is stored in its own file.
52 The objects are splayed over 256 subdirectories using
53 the first two characters of the sha1 object name to
54 keep the number of directory entries in `objects`
55 itself to a manageable number. Objects found
56 here are often called 'unpacked' (or 'loose') objects.
59 Packs (files that store many object in compressed form,
60 along with index files to allow them to be randomly
61 accessed) are found in this directory.
64 Additional information about the object store is
65 recorded in this directory.
68 This file is to help dumb transports discover what packs
69 are available in this object store. Whenever a pack is
70 added or removed, `git update-server-info` should be run
71 to keep this file up-to-date if the repository is
72 published for dumb transports. 'git repack' does this
75 objects/info/alternates::
76 This file records paths to alternate object stores that
77 this object store borrows objects from, one pathname per
78 line. Note that not only native Git tools use it locally,
79 but the HTTP fetcher also tries to use it remotely; this
80 will usually work if you have relative paths (relative
81 to the object database, not to the repository!) in your
82 alternates file, but it will not work if you use absolute
83 paths unless the absolute path in filesystem and web URL
84 is the same. See also 'objects/info/http-alternates'.
86 objects/info/http-alternates::
87 This file records URLs to alternate object stores that
88 this object store borrows objects from, to be used when
89 the repository is fetched over HTTP.
92 References are stored in subdirectories of this
93 directory. The 'git prune' command knows to preserve
94 objects reachable from refs found in this directory and
98 records tip-of-the-tree commit objects of branch `name`
101 records any object name (not necessarily a commit
102 object, or a tag object that points at a commit object).
104 refs/remotes/`name`::
105 records tip-of-the-tree commit objects of branches copied
106 from a remote repository.
108 refs/replace/`<obj-sha1>`::
109 records the SHA-1 of the object that replaces `<obj-sha1>`.
110 This is similar to info/grafts and is internally used and
111 maintained by linkgit:git-replace[1]. Such refs can be exchanged
112 between repositories while grafts are not.
115 records the same information as refs/heads/, refs/tags/,
116 and friends record in a more efficient way. See
117 linkgit:git-pack-refs[1].
120 A symref (see glossary) to the `refs/heads/` namespace
121 describing the currently active branch. It does not mean
122 much if the repository is not associated with any working tree
123 (i.e. a 'bare' repository), but a valid Git repository
124 *must* have the HEAD file; some porcelains may use it to
125 guess the designated "default" branch of the repository
126 (usually 'master'). It is legal if the named branch
127 'name' does not (yet) exist. In some legacy setups, it is
128 a symbolic link instead of a symref that points at the current
131 HEAD can also record a specific commit directly, instead of
132 being a symref to point at the current branch. Such a state
133 is often called 'detached HEAD.' See linkgit:git-checkout[1]
137 A slightly deprecated way to store shorthands to be used
138 to specify a URL to 'git fetch', 'git pull' and 'git push'.
139 A file can be stored as `branches/<name>` and then
140 'name' can be given to these commands in place of
141 'repository' argument. See the REMOTES section in
142 linkgit:git-fetch[1] for details. This mechanism is legacy
143 and not likely to be found in modern repositories.
146 Hooks are customization scripts used by various Git
147 commands. A handful of sample hooks are installed when
148 'git init' is run, but all of them are disabled by
149 default. To enable, the `.sample` suffix has to be
150 removed from the filename by renaming.
151 Read linkgit:githooks[5] for more details about
155 The current index file for the repository. It is
156 usually not found in a bare repository.
159 Additional information about the repository is recorded
163 This file helps dumb transports discover what refs are
164 available in this repository. If the repository is
165 published for dumb transports, this file should be
166 regenerated by 'git update-server-info' every time a tag
167 or branch is created or modified. This is normally done
168 from the `hooks/update` hook, which is run by the
169 'git-receive-pack' command when you 'git push' into the
173 This file records fake commit ancestry information, to
174 pretend the set of parents a commit has is different
175 from how the commit was actually created. One record
176 per line describes a commit and its fake parents by
177 listing their 40-byte hexadecimal object names separated
178 by a space and terminated by a newline.
181 This file, by convention among Porcelains, stores the
182 exclude pattern list. `.gitignore` is the per-directory
183 ignore file. 'git status', 'git add', 'git rm' and
184 'git clean' look at it but the core Git commands do not look
185 at it. See also: linkgit:gitignore[5].
188 Stores shorthands for URL and default refnames for use
189 when interacting with remote repositories via 'git fetch',
190 'git pull' and 'git push' commands. See the REMOTES section
191 in linkgit:git-fetch[1] for details. This mechanism is legacy
192 and not likely to be found in modern repositories.
195 Records of changes made to refs are stored in this
196 directory. See linkgit:git-update-ref[1]
197 for more information.
199 logs/refs/heads/`name`::
200 Records all changes made to the branch tip named `name`.
202 logs/refs/tags/`name`::
203 Records all changes made to the tag named `name`.
206 This is similar to `info/grafts` but is internally used
207 and maintained by shallow clone mechanism. See `--depth`
208 option to linkgit:git-clone[1] and linkgit:git-fetch[1].
213 linkgit:git-clone[1],
214 linkgit:git-fetch[1],
215 linkgit:git-pack-refs[1],
217 linkgit:git-checkout[1],
218 linkgit:gitglossary[7],
219 link:user-manual.html[The Git User's Manual]
223 Part of the linkgit:git[1] suite.