Package org.apache.commons.compress.archivers.dump

This package provides stream classes for reading archives using the Unix DUMP format. This format is similar to (and contemporary with) TAR but reads the raw filesystem directly. This means that writers are filesystem-specific even though the created archives are filesystem-agnostic.

Unlike other formats DUMP offers clean support for sparse files, extended attributes, and other file metadata. In addition DUMP supports incremental dump files can capture (most) file deletion. It also provides a native form of compression and will soon support native encryption as well.

In practice TAR archives are used for both distribution and backups. DUMP archives are used exclusively for backups.

Like any 30+-year-old application there are a number of variants. For pragmatic reasons we will only support archives with the 'new' tape header and inode formats. Other restrictions:

  • We only support ZLIB compression. The format also permits LZO and BZLIB compression.
  • Sparse files will have the holes filled.
  • MacOS finder and resource streams are ignored.
  • Extended attributes are not currently provided.
  • SELinux labels are not currently provided.

As of Apache Commons Compress 1.3 support for the dump format is read-only.