@c This is part of the paxutils manual.
@c Copyright (C) 2005, 2007, 2014 Free Software Foundation, Inc.
@c Written by Sergey Poznyakoff
@c This file is distributed under GFDL 1.1 or any later version
@c published by the Free Software Foundation.

  A @dfn{snapshot file} (or @dfn{directory file}) is created during
incremental backups (@pxref{Incremental Dumps}).  It
contains the status of the file system at the time of the dump and is
used to determine which files were modified since the last backup.

  @GNUTAR{} version @value{VERSION} supports three snapshot file
formats.  The first format, called @dfn{format 0}, is the one used by
@GNUTAR{} versions up to and including 1.15.1. The second format, called 
@dfn{format 1} is an extended version of this format, that contains more
metadata and allows for further extensions. It was used by alpha release
version 1.15.90. For alpha version 1.15.91 and stable releases
version 1.16 up through @value{VERSION}, the @dfn{format 2} is used.

  @GNUTAR{} is able to read all three formats, but will create
snapshots only in format 2.

  This appendix describes all three formats in detail.

@enumerate 0
@cindex format 0, snapshot file
@cindex snapshot file, format 0
@item
  @samp{Format 0} snapshot file begins with a line containing a
decimal number that represents a @acronym{UNIX} timestamp of the
beginning of the last archivation. This line is followed by directory
metadata descriptions, one per line. Each description has the
following format:

@smallexample
[@var{nfs}]@var{dev} @var{inode} @var{name}
@end smallexample

@noindent
where:

@table @var
@item nfs
A single plus character (@samp{+}), if this directory is located on
an @acronym{NFS}-mounted partition, otherwise empty.  

(That is, for non-NFS directories, the first character on the
description line contains the start of the @var{dev} field.)

@item dev
Device number of the directory;

@item inode
I-node number of the directory;

@item name
Name of the directory. Any special characters (white-space,
backslashes, etc.) are quoted.
@end table

@cindex format 1, snapshot file
@cindex snapshot file, format 1
@item
  @samp{Format 1} snapshot file begins with a line specifying the
format of the file. This line has the following structure:

@smallexample
@samp{GNU tar-}@var{tar-version}@samp{-}@var{incr-format-version}
@end smallexample

@noindent
where @var{tar-version} is the version number of @GNUTAR{}
implementation that created this snapshot, and
@var{incr-format-version} is the version number of the snapshot format
(in this case @samp{1}).

  Next line contains two decimal numbers, representing the
time of the last backup. First number is the number of seconds, the
second one is the number of nanoseconds, since the beginning of the
epoch.

  Lines that follow contain directory metadata, one line per
directory. Each line is formatted as follows:

@smallexample
[@var{nfs}]@var{mtime-sec} @var{mtime-nsec} @var{dev} @var{inode} @var{name}
@end smallexample

@noindent
where @var{mtime-sec} and @var{mtime-nsec} represent last
modification time of this directory with nanosecond precision;
@var{nfs}, @var{dev}, @var{inode} and @var{name} have the same meaning
as with @samp{format 0}.

@cindex format 2, snapshot file
@cindex snapshot file, format 2
@item
  @samp{Format 2} snapshot file begins with a format identifier, as described for
version 1, e.g.:

@smallexample
GNU tar-@value{VERSION}-2
@end smallexample

  This line is followed by newline. Rest of file consists of
records, separated by null (@acronym{ASCII} 0)
characters. Thus, in contrast to the previous formats, format 2
snapshot is a binary file.

  First two records are decimal integers, representing the
time of the last backup.  First number is the number of seconds, the
second one is the number of nanoseconds, since the beginning of the
epoch.  These are followed by arbitrary number of directory records.

  Each @dfn{directory record} contains a set of metadata describing a
particular directory.  Parts of a directory record are delimited with
@acronym{ASCII} 0 characters.  The following table describes each
part.  The @dfn{Number} type in this table stands for a decimal integer
in @acronym{ASCII} notation.  (Negative values are preceded with a "-"
character, while positive values have no leading punctuation.)

@multitable @columnfractions 0.25 0.15 0.6
@headitem Field @tab Type @tab Description
@item nfs @tab Character @tab @samp{1} if the directory is located on
an @acronym{NFS}-mounted partition, or @samp{0} otherwise;
@item timestamp_sec @tab Number @tab Modification time, seconds;
@item timestamp_nsec @tab Number @tab Modification time, nanoseconds;
@item dev @tab Number @tab Device number;
@item ino @tab Number @tab I-node number;
@item name @tab String @tab Directory name; in contrast to the
previous versions it is not quoted;
@item contents @tab Dumpdir @tab Contents of the directory;
@xref{Dumpdir}, for a description of its format.
@item
@end multitable

  Dumpdirs stored in snapshot files contain only records of types
@samp{Y}, @samp{N} and @samp{D}.

@cindex snapshot file field ranges
@opindex show-snapshot-field-ranges
The specific range of values allowed in each of the @dfn{Number} fields
depends on the underlying C datatypes as determined when @command{tar}
is compiled.  To see the specific ranges allowed for a particular
@command{tar} binary, you can use the
@option{--show-snapshot-field-ranges} option:

@smallexample
$ @kbd{tar --show-shapshot-field-ranges}
This tar's snapshot file field ranges are
   (field name      => [ min, max ]):

    nfs             => [ 0, 1 ],
    timestamp_sec   => [ -9223372036854775808, 9223372036854775807 ],
    timestamp_nsec  => [ 0, 999999999 ],
    dev             => [ 0, 18446744073709551615 ],
    ino             => [ 0, 18446744073709551615 ],
@end smallexample

(This example is from a GNU/Linux x86_64 system.)

@end enumerate

@c End of snapshot.texi