Changelog
xz-utils (4.999.9beta+20100117-1) unstable; urgency=low
* New upstream snapshot, taken from upstream commit 07a11dad.
+ The user-specified memory usage limit was not being honored in
some cases.
+ xz will no longer compress or decompress special files in place,
even with --force.
--force still enables compression of the target of a symlink.
+ xz will no longer read compressed data from a terminal or write
it to a terminal, even with --force.
+ liblzma: many Index handling API improvements:
- An lzma_index includes more useful metadata that is not
actually from the Index. This is particularly helpful for
describing multiple-stream files.
* lzma_index_cat() does not take a stream padding argument
any more; use lzma_index_stream_padding() instead.
* lzma_index_memusage() takes two arguments now.
The new lzma_index_memused() convenience function
calculates the total memory usage of an lzma_index.
* lzma_index_count() has been replaced with
lzma_index_block_count() and lzma_index_stream_count().
* The new lzma_index_stream_flags() allows storing Stream
Flags (e.g., checksum types) with the Index information.
* The new lzma_index_checks() function indicates the list of
checksum types in use.
* lzma_index_equal() has been removed. Its semantics were
unclear, and something approximate equivalent can be
implemented by iterating over Blocks in application code.
- An lzma_index no longer represents a position within an Index.
The new lzma_index_iter type can be used for that purpose.
* Certain functions generate an iterator on demand now rather
than clobbering their lzma_index argument. Using a single
lzma_index for multiple concurrent invocations is safe now:
lzma_index_buffer_encode(), lzma_index_encoder()
* Functions that used to specify a Record by modifying an
lzma_index argument have been renamed and taught to write
an lzma_index_iter instead:
lzma_index_locate() -> lzma_index_iter_locate()
lzma_index_rewind() -> lzma_index_iter_rewind()
lzma_index_read() -> lzma_index_iter_next()
* The lzma_index_record type used by lzma_index_read() for
complex seeks is no more. lzma_index_iter_next() instead
takes a parameter of the new lzma_index_iter_mode type.
* New function: lzma_index_iter_init()
* lzma_index_init() cannot reuse memory from an existing
lzma_index any more. The signature changed accordingly.
+ xz: support for decompressing to a sparse file
- Try to write a sparse file if the decompressed stream contains
long runs of null bytes.
- The new --no-sparse option disables this behavior.
+ xz: add Czech translation.
- liblzma-dev: lzma_filters_copy() and lzma_filters_update() are
declared not to throw exceptions.
* liblzma: bump soname.
* xz-utils: Add NEWS.Debian describing the sparse file support.
* debian/rules: Stop explicitly disabling assembler optimizations on
the Hurd. The configure script will check if a platform is
supported.
* debian/rules: Update clean target to handle comments and wildcards
in .gitignore.
* Build-Depends: perl.
* debian/rules: Speed up build by using autoreconf --symlink.
* Remove debian/dirs lists.
* Update copyright file and fix a small typo.
-- Jonathan Nieder <email address hidden> Mon, 18 Jan 2010 04:19:50 -0600
