| blob | 74cc7c1cb831c700c14406f6e306e7a3002054d2 |
1 git-pack-objects(1)
2 ===================
4 NAME
5 ----
6 git-pack-objects - Create a packed archive of objects
9 SYNOPSIS
10 --------
11 [verse]
12 'git-pack-objects' [-q] [--no-reuse-delta] [--delta-base-offset] [--non-empty]
13 [--local] [--incremental] [--window=N] [--depth=N] [--all-progress]
14 [--revs [--unpacked | --all]*] [--stdout | base-name] < object-list
17 DESCRIPTION
18 -----------
19 Reads list of objects from the standard input, and writes a packed
20 archive with specified base-name, or to the standard output.
22 A packed archive is an efficient way to transfer set of objects
23 between two repositories, and also is an archival format which
24 is efficient to access. The packed archive format (.pack) is
25 designed to be unpackable without having anything else, but for
26 random access, accompanied with the pack index file (.idx).
28 Placing both in the pack/ subdirectory of $GIT_OBJECT_DIRECTORY (or
29 any of the directories on $GIT_ALTERNATE_OBJECT_DIRECTORIES)
30 enables git to read from such an archive.
32 'git-unpack-objects' command can read the packed archive and
33 expand the objects contained in the pack into "one-file
34 one-object" format; this is typically done by the smart-pull
35 commands when a pack is created on-the-fly for efficient network
36 transport by their peers.
38 In a packed archive, an object is either stored as a compressed
39 whole, or as a difference from some other object. The latter is
40 often called a delta.
43 OPTIONS
44 -------
45 base-name::
46 Write into a pair of files (.pack and .idx), using
47 <base-name> to determine the name of the created file.
48 When this option is used, the two files are written in
49 <base-name>-<SHA1>.{pack,idx} files. <SHA1> is a hash
50 of the sorted object names to make the resulting filename
51 based on the pack content, and written to the standard
52 output of the command.
54 --stdout::
55 Write the pack contents (what would have been written to
56 .pack file) out to the standard output.
58 --revs::
59 Read the revision arguments from the standard input, instead of
60 individual object names. The revision arguments are processed
61 the same way as linkgit:git-rev-list[1] with `--objects` flag
62 uses its `commit` arguments to build the list of objects it
63 outputs. The objects on the resulting list are packed.
65 --unpacked::
66 This implies `--revs`. When processing the list of
67 revision arguments read from the standard input, limit
68 the objects packed to those that are not already packed.
70 --all::
71 This implies `--revs`. In addition to the list of
72 revision arguments read from the standard input, pretend
73 as if all refs under `$GIT_DIR/refs` are specified to be
74 included.
76 --window=[N], --depth=[N]::
77 These two options affect how the objects contained in
78 the pack are stored using delta compression. The
79 objects are first internally sorted by type, size and
80 optionally names and compared against the other objects
81 within --window to see if using delta compression saves
82 space. --depth limits the maximum delta depth; making
83 it too deep affects the performance on the unpacker
84 side, because delta data needs to be applied that many
85 times to get to the necessary object.
86 The default value for --window is 10 and --depth is 50.
88 --window-memory=[N]::
89 This option provides an additional limit on top of `--window`;
90 the window size will dynamically scale down so as to not take
91 up more than N bytes in memory. This is useful in
92 repositories with a mix of large and small objects to not run
93 out of memory with a large window, but still be able to take
94 advantage of the large window for the smaller objects. The
95 size can be suffixed with "k", "m", or "g".
96 `--window-memory=0` makes memory usage unlimited, which is the
97 default.
99 --max-pack-size=<n>::
100 Maximum size of each output packfile, expressed in MiB.
101 If specified, multiple packfiles may be created.
102 The default is unlimited.
104 --incremental::
105 This flag causes an object already in a pack ignored
106 even if it appears in the standard input.
108 --local::
109 This flag is similar to `--incremental`; instead of
110 ignoring all packed objects, it only ignores objects
111 that are packed and not in the local object store
112 (i.e. borrowed from an alternate).
114 --non-empty::
115 Only create a packed archive if it would contain at
116 least one object.
118 --progress::
119 Progress status is reported on the standard error stream
120 by default when it is attached to a terminal, unless -q
121 is specified. This flag forces progress status even if
122 the standard error stream is not directed to a terminal.
124 --all-progress::
125 When --stdout is specified then progress report is
126 displayed during the object count and deltification phases
127 but inhibited during the write-out phase. The reason is
128 that in some cases the output stream is directly linked
129 to another command which may wish to display progress
130 status of its own as it processes incoming pack data.
131 This flag is like --progress except that it forces progress
132 report for the write-out phase as well even if --stdout is
133 used.
135 -q::
136 This flag makes the command not to report its progress
137 on the standard error stream.
139 --no-reuse-delta::
140 When creating a packed archive in a repository that
141 has existing packs, the command reuses existing deltas.
142 This sometimes results in a slightly suboptimal pack.
143 This flag tells the command not to reuse existing deltas
144 but compute them from scratch.
146 --no-reuse-object::
147 This flag tells the command not to reuse existing object data at all,
148 including non deltified object, forcing recompression of everything.
149 This implies --no-reuse-delta. Useful only in the obscure case where
150 wholesale enforcement of a different compression level on the
151 packed data is desired.
153 --compression=[N]::
154 Specifies compression level for newly-compressed data in the
155 generated pack. If not specified, pack compression level is
156 determined first by pack.compression, then by core.compression,
157 and defaults to -1, the zlib default, if neither is set.
158 Add \--no-reuse-object if you want to force a uniform compression
159 level on all data no matter the source.
161 --delta-base-offset::
162 A packed archive can express base object of a delta as
163 either 20-byte object name or as an offset in the
164 stream, but older version of git does not understand the
165 latter. By default, git-pack-objects only uses the
166 former format for better compatibility. This option
167 allows the command to use the latter format for
168 compactness. Depending on the average delta chain
169 length, this option typically shrinks the resulting
170 packfile by 3-5 per-cent.
172 --threads=<n>::
173 Specifies the number of threads to spawn when searching for best
174 delta matches. This requires that pack-objects be compiled with
175 pthreads otherwise this option is ignored with a warning.
176 This is meant to reduce packing time on multiprocessor machines.
177 The required amount of memory for the delta search window is
178 however multiplied by the number of threads.
180 --index-version=<version>[,<offset>]::
181 This is intended to be used by the test suite only. It allows
182 to force the version for the generated pack index, and to force
183 64-bit index entries on objects located above the given offset.
186 Author
187 ------
188 Written by Linus Torvalds <torvalds@osdl.org>
190 Documentation
191 -------------
192 Documentation by Junio C Hamano
194 See Also
195 --------
196 linkgit:git-rev-list[1]
197 linkgit:git-repack[1]
198 linkgit:git-prune-packed[1]
200 GIT
201 ---
202 Part of the linkgit:git[7] suite