Documentation / git-pack-objects.txton commit Merge branch 'rs/git-dir-cleanup' (263830c)
   1git-pack-objects(1)
   2===================
   3
   4NAME
   5----
   6git-pack-objects - Create a packed archive of objects
   7
   8
   9SYNOPSIS
  10--------
  11[verse]
  12'git pack-objects' [-q | --progress | --all-progress] [--all-progress-implied]
  13        [--no-reuse-delta] [--delta-base-offset] [--non-empty]
  14        [--local] [--incremental] [--window=N] [--depth=N]
  15        [--revs [--unpacked | --all]*] [--stdout | base-name]
  16        [--keep-true-parents] < object-list
  17
  18
  19DESCRIPTION
  20-----------
  21Reads list of objects from the standard input, and writes a packed
  22archive with specified base-name, or to the standard output.
  23
  24A packed archive is an efficient way to transfer set of objects
  25between two repositories, and also is an archival format which
  26is efficient to access.  The packed archive format (.pack) is
  27designed to be self contained so that it can be unpacked without
  28any further information, but for fast, random access to the objects
  29in the pack, a pack index file (.idx) will be generated.
  30
  31Placing both in the pack/ subdirectory of $GIT_OBJECT_DIRECTORY (or
  32any of the directories on $GIT_ALTERNATE_OBJECT_DIRECTORIES)
  33enables git to read from such an archive.
  34
  35The 'git unpack-objects' command can read the packed archive and
  36expand the objects contained in the pack into "one-file
  37one-object" format; this is typically done by the smart-pull
  38commands when a pack is created on-the-fly for efficient network
  39transport by their peers.
  40
  41In a packed archive, an object is either stored as a compressed
  42whole, or as a difference from some other object.  The latter is
  43often called a delta.
  44
  45
  46OPTIONS
  47-------
  48base-name::
  49        Write into a pair of files (.pack and .idx), using
  50        <base-name> to determine the name of the created file.
  51        When this option is used, the two files are written in
  52        <base-name>-<SHA1>.{pack,idx} files.  <SHA1> is a hash
  53        of the sorted object names to make the resulting filename
  54        based on the pack content, and written to the standard
  55        output of the command.
  56
  57--stdout::
  58        Write the pack contents (what would have been written to
  59        .pack file) out to the standard output.
  60
  61--revs::
  62        Read the revision arguments from the standard input, instead of
  63        individual object names.  The revision arguments are processed
  64        the same way as 'git rev-list' with the `--objects` flag
  65        uses its `commit` arguments to build the list of objects it
  66        outputs.  The objects on the resulting list are packed.
  67
  68--unpacked::
  69        This implies `--revs`.  When processing the list of
  70        revision arguments read from the standard input, limit
  71        the objects packed to those that are not already packed.
  72
  73--all::
  74        This implies `--revs`.  In addition to the list of
  75        revision arguments read from the standard input, pretend
  76        as if all refs under `$GIT_DIR/refs` are specified to be
  77        included.
  78
  79--include-tag::
  80        Include unasked-for annotated tags if the object they
  81        reference was included in the resulting packfile.  This
  82        can be useful to send new tags to native git clients.
  83
  84--window=[N]::
  85--depth=[N]::
  86        These two options affect how the objects contained in
  87        the pack are stored using delta compression.  The
  88        objects are first internally sorted by type, size and
  89        optionally names and compared against the other objects
  90        within --window to see if using delta compression saves
  91        space.  --depth limits the maximum delta depth; making
  92        it too deep affects the performance on the unpacker
  93        side, because delta data needs to be applied that many
  94        times to get to the necessary object.
  95        The default value for --window is 10 and --depth is 50.
  96
  97--window-memory=[N]::
  98        This option provides an additional limit on top of `--window`;
  99        the window size will dynamically scale down so as to not take
 100        up more than N bytes in memory.  This is useful in
 101        repositories with a mix of large and small objects to not run
 102        out of memory with a large window, but still be able to take
 103        advantage of the large window for the smaller objects.  The
 104        size can be suffixed with "k", "m", or "g".
 105        `--window-memory=0` makes memory usage unlimited, which is the
 106        default.
 107
 108--max-pack-size=[N]::
 109        Maximum size of each output pack file. The size can be suffixed with
 110        "k", "m", or "g". The minimum size allowed is limited to 1 MiB.
 111        If specified,  multiple packfiles may be created.
 112        The default is unlimited, unless the config variable
 113        `pack.packSizeLimit` is set.
 114
 115--honor-pack-keep::
 116        This flag causes an object already in a local pack that
 117        has a .keep file to be ignored, even if it appears in the
 118        standard input.
 119
 120--incremental::
 121        This flag causes an object already in a pack ignored
 122        even if it appears in the standard input.
 123
 124--local::
 125        This flag is similar to `--incremental`; instead of
 126        ignoring all packed objects, it only ignores objects
 127        that are packed and/or not in the local object store
 128        (i.e. borrowed from an alternate).
 129
 130--non-empty::
 131        Only create a packed archive if it would contain at
 132        least one object.
 133
 134--progress::
 135        Progress status is reported on the standard error stream
 136        by default when it is attached to a terminal, unless -q
 137        is specified. This flag forces progress status even if
 138        the standard error stream is not directed to a terminal.
 139
 140--all-progress::
 141        When --stdout is specified then progress report is
 142        displayed during the object count and compression phases
 143        but inhibited during the write-out phase. The reason is
 144        that in some cases the output stream is directly linked
 145        to another command which may wish to display progress
 146        status of its own as it processes incoming pack data.
 147        This flag is like --progress except that it forces progress
 148        report for the write-out phase as well even if --stdout is
 149        used.
 150
 151--all-progress-implied::
 152        This is used to imply --all-progress whenever progress display
 153        is activated.  Unlike --all-progress this flag doesn't actually
 154        force any progress display by itself.
 155
 156-q::
 157        This flag makes the command not to report its progress
 158        on the standard error stream.
 159
 160--no-reuse-delta::
 161        When creating a packed archive in a repository that
 162        has existing packs, the command reuses existing deltas.
 163        This sometimes results in a slightly suboptimal pack.
 164        This flag tells the command not to reuse existing deltas
 165        but compute them from scratch.
 166
 167--no-reuse-object::
 168        This flag tells the command not to reuse existing object data at all,
 169        including non deltified object, forcing recompression of everything.
 170        This implies --no-reuse-delta. Useful only in the obscure case where
 171        wholesale enforcement of a different compression level on the
 172        packed data is desired.
 173
 174--compression=[N]::
 175        Specifies compression level for newly-compressed data in the
 176        generated pack.  If not specified,  pack compression level is
 177        determined first by pack.compression,  then by core.compression,
 178        and defaults to -1,  the zlib default,  if neither is set.
 179        Add --no-reuse-object if you want to force a uniform compression
 180        level on all data no matter the source.
 181
 182--delta-base-offset::
 183        A packed archive can express base object of a delta as
 184        either 20-byte object name or as an offset in the
 185        stream, but older version of git does not understand the
 186        latter.  By default, 'git pack-objects' only uses the
 187        former format for better compatibility.  This option
 188        allows the command to use the latter format for
 189        compactness.  Depending on the average delta chain
 190        length, this option typically shrinks the resulting
 191        packfile by 3-5 per-cent.
 192
 193--threads=<n>::
 194        Specifies the number of threads to spawn when searching for best
 195        delta matches.  This requires that pack-objects be compiled with
 196        pthreads otherwise this option is ignored with a warning.
 197        This is meant to reduce packing time on multiprocessor machines.
 198        The required amount of memory for the delta search window is
 199        however multiplied by the number of threads.
 200        Specifying 0 will cause git to auto-detect the number of CPU's
 201        and set the number of threads accordingly.
 202
 203--index-version=<version>[,<offset>]::
 204        This is intended to be used by the test suite only. It allows
 205        to force the version for the generated pack index, and to force
 206        64-bit index entries on objects located above the given offset.
 207
 208--keep-true-parents::
 209        With this option, parents that are hidden by grafts are packed
 210        nevertheless.
 211
 212
 213Author
 214------
 215Written by Linus Torvalds <torvalds@osdl.org>
 216
 217Documentation
 218-------------
 219Documentation by Junio C Hamano
 220
 221SEE ALSO
 222--------
 223linkgit:git-rev-list[1]
 224linkgit:git-repack[1]
 225linkgit:git-prune-packed[1]
 226
 227GIT
 228---
 229Part of the linkgit:git[1] suite