Update readme

README.md

@@ -1,26 +1,29 @@
- The __Fast LZMA2 Library__ is a lossless high-ratio data compression library based on the LZMA2 codec in 7-zip.
+The __Fast LZMA2 Library__ is a lossless high-ratio data compression library based on the LZMA2 codec in 7-zip.
 
-Binaries of 7-Zip forks which use the algorithm are available in the [7-Zip-FL2 project] and the [7-Zip-zstd project].
+Binaries of 7-Zip forks which use the algorithm are available in the [7-Zip-FL2 project] and the [7-Zip-zstd project]. The library
+is also embedded in a fork of XZ Utils, named [FXZ Utils].
 
 [7-Zip-FL2 project]: https://github.com/conor42/7-Zip-FL2/releases/
 [7-Zip-zstd project]: https://github.com/mcmilk/7-Zip-zstd/releases/
+[FXZ Utils]: https://github.com/conor42/fxz
 
-The library uses a parallel buffered radix match finder and some optimizations from Zstandard to achieve a 20% to 100%
+The library uses a parallel buffered radix match-finder and some optimizations from Zstandard to achieve a 20% to 100%
 speed gain at the higher levels over the default LZMA2 algorithm used in 7-zip, for a small loss in compression ratio. Speed gains
 depend on the nature of the source data. The library also uses some threading, portability, and testing code from Zstandard.
 
-Use of the parallel buffered radix match finder allows multithreaded execution with a simple design and low memory requirement. The
-library can compress using many threads without dividing the input into large chunks requiring the duplication of the match
+Use of the radix match-finder allows multi-threaded execution employing a simple threading model and with low memory usage. The
+library can compress using many threads without dividing the input into large chunks which require the duplication of the match-
 finder tables and chains. Extra memory used per thread is typically no more than a few megabytes.
 
-The largest caveat is that the matchfinder is a block algorithm, and to achieve about the same ratio as 7-Zip requires double the
-dictionary size, which raises the decompression memory usage. By default it uses the same dictionary size as 7-Zip, resulting in a
-small loss of ratio. A high-compression option is provided to select parameters which achieve higher compression on smaller
-dictionaries. The speed/ratio tradeoff is less optimal with this enabled.
+The largest caveat is that the match-finder is a block algorithm, and to achieve about the same ratio as 7-Zip requires double the
+dictionary size, which raises the decompression memory usage. By default it uses the same dictionary size as 7-Zip, resulting in
+output that is larger by about 1% of the compressed size. A high-compression option is provided to select parameters which achieve
+higher compression on smaller dictionaries. The speed/ratio tradeoff is less optimal with this enabled.
+
+Here are the results of an in-memory benchmark using two threads on the [Silesia compression corpus] vs the 7-zip 19.00 LZMA2
+encoder. The design goal for the encoder and compression level parameters was to move the line as far as possible toward the top
+left of the graph. This provides an optimal speed/ratio tradeoff.
 
-Tested in memory vs the 7-zip 19.00 LZMA2 encoder on the [Silesia compression corpus] using two threads. The design goal for the
-encoder and compression level parameters was to move the line as far as possible toward the top left of the graph. This provides an
-optimal speed/ratio tradeoff.
 
 [Silesia compression corpus]: http://sun.aei.polsl.pl/~sdeor/index.php?page=silesia
 
@@ -31,31 +34,43 @@ Compression data rate vs ratio
 ### Build
 
 Build methods are not yet comprehensive for all systems. There are VS 2015 projects for building a benchmark program, fuzz tester,
-and a DLL. Makefiles for gcc are included for the benchmark, fuzzer and DLL, and user nemequ has contributed a CMake file. If anyone
-would like to help improve the build methods, please do so.
+and a DLL. Makefiles for gcc are included for the benchmark, fuzzer and DLL, and Evan Nemerson has contributed a CMake file, which
+currently does not enable the assembler-optimized decoder. If anyone would like to help improve the build methods, please do so.
+On POSIX systems, FXZ Utils provides a derivative of liblzma from XZ Utils as a wrapper for Fast LZMA2. It is built using GNU
+autotools.
 
 ### Status
 
 The library has passed a large amount of fuzz testing, and testing on file sets selected at random in the Radyx file archiver. An
-earlier version was released in the 7-Zip forks linked above. The library is suitable for production environments.
+earlier version was released in the 7-Zip forks linked above. The library is considered suitable for production environments.
+However, no warranty or fitness for a particular purpose is expressed or implied.
+
 
 Changes in v1.0.0:
 
-- Redesigned API.
-- Moved detection of repeats from the single-threaded initialization stage to a later, multithreaded stage.
-- LZMA2 encoder optimizations; some from Igor Pavlov's optimizations in 7-Zip 18.05, and some additional ones.
+- Breaking changes have been made to the API functions.
+- Some of the options have been renamed.
+- Optimized the encoder, incorporating some of Igor Pavlov's improvements to 7-Zip 18.05, and some novel ones. The speed increase is
+  about 5% - 8%.
+- Moved detection of repeats from the single-threaded initialization stage to a later, multi-threaded stage to increase speed.
 - Removed two compression levels, reducing the total to 10, and tweaked the parameters.
+- Improved calculation of the match buffer size. It can still be changed in the options, but the meaning of the value is different.
+- Replaced the callbacks for writing and progress with timeouts and new functions to gain direct access to the dictionary buffer and
+  the compressed data buffers.
+- Added Igor Pavlov's assembler-optimized decoder.
 - Multithreaded decompression.
 
+
 Changes in v0.9.2:
 
 - Fixed excess memory allocation when the dictionary size is > 64Mb
 
+
 Changes in v0.9.1:
 
 - Fixed a bug in compression of very small files when using a high search depth.
-- Added an incompressibility checker which processes high-entropy (e.g. encrypted or
-  already compressed) data about twice as fast as before.
+- Added an incompressibility checker which processes high-entropy (e.g. encrypted or already compressed) data about twice as fast
+  as before.
 
 ### License