Skip to content
Tutorials Center
My learning

Topics

Git Reference ยท Plumbing Commands

git unpack-objects command

Reference for git unpack-objects: read a packfile from standard input and write its objects as loose files, with -n, -q, --strict and size-limit notes.

git unpack-objects is the reverse of packing. It reads a packfile from standard input and writes every object in it as an individual loose file under .git/objects/xx/. It does not keep the pack, and it does not need an index.

Git uses it when a small amount of data arrives over the network: it is cheaper to unpack a handful of objects than to store a tiny pack. You can use it directly to import a pack produced by git pack-objects --stdout, or to see what a pack really contains.

Terminal
git unpack-objects [-n] [-q] [--strict] < file.pack

Try it

Bash Copy objects between repositories

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with read or cat), it gets the next line from this box.

Bash Dry run and repeated input

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with read or cat), it gets the next line from this box.

Syntax

Terminal
git unpack-objects [-n] [-q] [-r] [--strict] [--max-input-size=<size>] < <pack-file>
Name Description
-n Dry run. Check the pack but write nothing.
-q Quiet: no progress output.
-r Recover: try to unpack as much as possible from a damaged pack instead of stopping at the first error.
--strict Check each object for validity and that its links lead to objects that exist or are in the same pack. Objects that fail are not written.
--max-input-size=<size> Refuse a pack larger than this size.
--pack_header=<version>,<count> Used internally when the pack header was already consumed by the caller.

Input: the pack must be on standard input; the command never reads a file by name. Nothing is printed on success.

Errors: a bad pack signature, a checksum mismatch or a truncated stream ends with fatal: and a nonzero status. Objects unpacked before the failure remain on disk.

Usage notes

  • Existing objects are skipped. Running the command twice with the same pack changes nothing the second time, because Git does not rewrite objects that already exist (loose or in a pack).
  • Deltas are resolved. A pack stores many objects as deltas; the loose files contain the full, zlib-compressed objects, so unpacking can use much more disk space than the pack did. A repository of 1 MB packed may need many megabytes loose.
  • Standard input only. The command reads a pack stream, not a file name or a directory. To expand a pack that already lives in a repository, copy the .pack file out and feed it to a different, empty repository; in the original repository Git would skip every object it already has.
  • Thin packs fail when the missing delta bases are not in the repository. Use git index-pack with --fix-thin for those.
  • Transfer limit. git fetch and git receive-pack call unpack-objects when a pack holds fewer than transfer.unpackLimit objects (default 100, or fetch.unpackLimit / receive.unpackLimit) and index-pack for larger ones.
  • Many loose files are slow. Run git gc or git repack afterwards if you unpacked a lot.