Quex 0.67.3-5 Release Notes:
CHANGES: -- Removed Accumulator and PostCategorizer from the code generator.
-- Removed options for above.
-- Code for above remains present in 'code_base/extra'. That is,
they may still be used.
-- Renamed command line options '--codec*' to '--encoding*'
-- Lexeme converters have been refactor to work more generally.
-- Token class in C++, now also depends only on 'char' not 'std::string'.
The later has become a significant performance drain in some
-- Bug fixes: 298, 299, 300.
-- Cleaning up.
Quex 0.67.2 Release Notes:
-- 'print_this()' function has been completed so that state of all
components is printed properly.
-- Extract further dependency of 'converter' from lexical analyzer
Quex 0.67.1 Release Notes:
-- Buffer split upon include: Upon include the unused part
of an including buffer may be used for the content of the
=> Massive memory usage reduction in extensive include
-- Cleansening of the user API, three schemes:
(1) File name + converter (optional)
(2) Byte loader + converter (both optional)
(3) Memory chunk
Schemes are present for constructor, include-push, and reset.
-- Unit tests for all three API schemes.
Quex 0.66.1 to 0.66.5 Release Notes:
-- 'on_bad_lexatom' handler in modes
Codec errors can now be detected directly during lexical
analysis. No extra run on the input data is required.
-- 'skip_range', 'skip_nested_range', and 'indentation' allow
now arbitrary regular expressions as boarders.
-- 'gavager' and 'feeder' adapters for manual buffer filling.
By means of these adapters the manual buffer filling has been
tremendously simplified. Example usage is show in the demo
-- Introducing DFA Algebra. Consistent struction for 'union'
'intersection', and 'complement' of patterns.
-- Bug fixes on compiler warnings Ubuntu 64bit Linux.
-- Bug fixes with respect to converters using binary mode streams.
-- Adapted API for constructors, include_push, and reset to deal
with binary mode in 'std::istream' and 'std::wistream'.
-- All loop-functionality (indentation count, character skip,
range skip, and nested range skip) has been expressed in
terms of state machines only. String-base code generation in
this regard has been removed. Code-wise this has been a major
step towards uniformity.
-- Added serveral unit tests for skipping and indentation based
-- Addes massive unit tests for converters using binary streams.
7.12.2016 Frank-Rene Schaefer