Import libarchive-3.0.2.
*libarchive 3.0.2 released
*Various fixes merged from FreeBSD
*Symlink support in Zip reader and writer
*Robustness fixes to 7Zip reader
*libarchive 3.0.1b released
*7Zip reader
*Small fixes to ISO and Zip to improve robustness with corrupted input
*Improve streaming Zip reader's support for uncompressed entries
*New seeking Zip reader supports SFX Zip archives
*libarchive 3.0.0a released
*Update shared-library version calculations for libarchive 3.x
*Fix tar -s; follow GNU tar for controlling hardlink/symlink substitutions
*Fix reading ISO images built by NetBSD's mkisofs
*Old archive_read_support_compression_XXX functions are deprecated and
will disappear in libarchive 4.0.
*RAR reader
*Add tar:compat-2x option to emulate broken libarchive 2.x
handling of pax UTF-8 headers
*Refactor read_open() into a collection of single-item setters;
support the old interfaces as wrappers
*Split disk writer into separate POSIX and Windows implementations
*More work to return errors instead of calling abort()
*Add charset option to many writers to control MBCS filenames
*Overhauled support for per-format extension options
*Track character set used for mbcs strings, support
translating to/from user-specified locale
*Recognize mtree files without requiring a signature
*Use iconv to convert to/from Unicode instead of making bad
assumptions about the C90 character set translation functions
*CAB reader
*LHA/LZH reader
*Many improvements to ISO reader compatibility
*Use larger buffers when copy files into an archive
*archive_read_disk now supports traversals
*XAR writer
*Fix ^T handling; don't exit on interrupted reads and writes
*Improved detection of platform-specific crypto support
*lzip read and write filters
*tar --gid --gname --uid --uname
*Use Red-black tree for ISO reader/writer to improve performance
*Minimal writer for legacy GNU tar format
*Relax handling of state failures; misuse by clients now generally
results in a sticky ARCHIVE_FATAL rather than a visit to abort()
*ISO writer
*Split many man pages into smaller chunks.
*Cheat on block sizes when reading archives from disk.
*Use int64_t instead of off_t, dev_t, ino_t, uid_t, and gid_t
*Document new ACL functions.
*Support multiple write filters
*Remove some legacy libarchive 1.x APIs
*Read afio headers
*Archive sparse files compatibly with GNU tar
*Support cpio -V