2.4.0 2015-06-21 * Fix: Added rules 21b and 22.02 introduced by Unicode 8.0.0. * Update -version-info from 4:6:1 to 4:7:1. 2.3.3 2015-06-21 * Withdrawn. * Update UCD to 8.0.0. 2.3.2 2014-06-20 * Update UCD to 7.0.0. 2.3.2beta1 2014-05-04 * Update UCD to 7.0.0 beta at 2014-03-18. * Moved repository to GitHub. 2.3.1 2013-09-28 * Update UCD to 6.3.0. * Typo in README.ja_JP. 2.3.1beta1 2013-03-26 * Update UCD to 6.3.0 beta at 2013-03-02. 2.3.0 2012-10-01 * Update UCD to 6.2.0. * data/custom.pl: * data/data.pl: * lib/charprop.c: * lib/gcstring.c: - EA_Z was devided to EA_Z, EA_ZA and EA_ZW which are nonspacing variants of EAW "N", "A" and "W", respectively. Currently, there are no differences between EA_Z and EA_ZA. - VIRAMA_AS_JOINER option had been applicable to versions prior to 6.1.0. 2.3.0gamma1 2012-08-18 * data/Rules-6.2.0beta.txt: * data/constants.pl: * data/custom.pl: * data/data.pl: * lib/charprop.c: * lib/gcstring.c: - Changes based on new drafts of UAX#14/UAX#29 (not yet published). GCB Joiner and After_Joiner and LBC ZJ are withdrawn by UC. As also NonJoiner. 2.3.0beta1 2012-07-14 * Update UCD to 6.2.0beta at Jun 05 2012. * config.guess: * config.sub: - updated. * data/constants.pl: * data/custom.pl: * data/data.pl: * data/makefile: - custom Grapheme_Cluster_Break property value "ZWJ" is divided to "NonJoiner" (U+200C ZERO WIDTH NON-JOINER) and "Joiner" (U+200D ZERO WIDTH JOINER). - Store custom GCB to *.custom file. * lib/charprop.c: - Added ZJ/Joiner case of guessed properties. * lib/gcstring.c: - Updated GCB rules. * Makefile.am: - increase -version-info from 4.5.1 to 4.6.1. 2.2.1 2012-04-11 * lib/gcstring.c: - Treat an isolated hangul jamo as wide, though some of them are assigned EAW neutral (N). cf. the discussion: http://lists.w3.org/Archives/Public/public-i18n-cjk/2011JanMar/thread.html#msg55 * REAME: * README.ja_JP: - About naming. 2.2.0 2012-04-02 * data/constants.pl: * data/data.pl: - NEW: added custom Grapheme_Cluster_Break property value "ZWJ", U+200C ZERO WIDTH JOINER or U+200D ZERO WIDTH NON-JOINER. They are lb:CM and gb:Extend but aren't gc:M so do not affect to Legacy-CM rule. * tests/01break.t: - Added Korean (NFD) test. * tests/03ns.t: - Added loose nonstarter test. 2.2.0beta 2012-03-30 * License Change: - From now on, sombok is distributed under dual license of GPL *and* Artistic License. * configure.ac: - Move variable version informations to files: UNICODE and VERSION. * lib/break.c: - New: linebreak_free_result(): to free result of linebreak_break*() so that implementation mismatches of malloc functions are hidden. * lib/charprop.c: - New: linebreak_merge_lbclass(), linebreak_merge_eawidth(). * lib/gcstring.c: - New: gcstring_lbclass(), gcstring_lbclass_ext(). * lib/linebreak.c: - Specify strict prototypes for some more functions. - linebreak_get_lbrule(): Concern LEGACY_CM and HANGUL_AS_AL options. - Obsoleted: linebreak_lbclass(), linebreak_lbclass_ext(), linebreak_eawidth(). * Makefile.am: - increase -version-info from 4.4.1 to 4.5.1. * Added TODO. 2.1.1 2012-02-25 * config.guess: * config.sub: - updated. * data/Rules-6.1.0.txt: - Correct revision date of UAX#14. * include/sombok.h.in: * lib/linebreak.c: - Specify strict prototypes of callback members of linebreak_t. * lib/break.c: * lib/gcstring.c: - Chg: Now Legacy-CM rule is applied during grapheme cluster breaking. If this rule is disabled, SPACE and succeeding combining mark are not conjoined. - Chg: Morbid hangul syllable block (e.g. <L Extend V T>) no longer does not form single grapheme cluster. 2.1.0 2012-02-04 * Release 2.1.0. * Makefile.am: - increase -version-info from 4.3.1 to 4.4.1. * tests/01break.t: - Added Sanskrit (sa) case. 2.1.0beta2 2012-01-29 * data/constants.pl: * data/data.pl: - NEW: added custom Grapheme_Cluster_Break property values "Virama" and "OtherLetter". Former means characters having canonical combining class 9 (Virama) (they are either Extend or SpacingMark). Latter means characters having general category L? and Grapheme_Cluster_Break Other. They are currently limited to Brahmic scripts. * lib/break.c: - NEW: linebreak_break_from_utf8(). - Chg: charmax may be 0 (unlimited). NOTE: this feature has not been tested enough. * lib/gcstring.c: - NEW: gcstring_new_from_utf8(). - NEW: LINEBREAK_OPTION_VIRAMA_AS_JOINER option of linebreak object controls if virama characters behave as consonant joiner. * lib/utf8.c: - NEW: sombok_decode_utf8(), sombok_encode_utf8(). 2.1.0beta 2012-01-01 * Bump up minor version. * Update UCD to 6.1.0beta at Dec 15 2011. * config.sub: * config.guess: - Updated. * data/Rules-6.0.0.txt: - Backport - removed LB25 entries. * lib/break.c: - _break_partial(): Fix: LB25 preprocessing can miss partial input. * lib/charprop.c: * lib/lineBreak.c: * include/sombok.h.in: - New option LINEBREAK_OPTION_NONSTARTER_LOOSE to control strictness of line breaking behavior. If it is set, a newly introduced class CJ will be treated as ID, otherwise NS. * t/02hangul.t: - Unicode 6.1.0beta changed general category of hangul bangjeom (U+302E and U+302F) from Mn to Mc. * [Note] ``Compact array'' became smaller than 8192 entries again. 2.0.6 2011-11-01 Backport from 2.1.0beta; Unicode data version 6.0.0. * data/Rule-6.0.0.txt: - Removed LB25 entries since they are handled by hardcoded preprocessing. * debian/*: - Updated Debian packaging info by which in Debian sid packaged by Emmanuel Bouthenot. * lib/break.c: - Implemented strict LB25 rule. See also UAX#14 8.2 Example 7. - Moved linebreak_lbrule() from linebreak.c to break.c. - For compatibility, linebreak_lbrule() gives results by simplified rules LB1, LB25 and emulates LB9, LB10. * src/sombok.c: - Fix SF.net bug #3400519(1): removed ESTRPIPE which is specific to Linux etc. Reported by Yusuke Muraoka (jbking@sf.net). * tests/01break.t: - Added Hebrew (he) test. * config.guess: * config.sub: - Updated. * COPYING: - Fix CPAN RT#71657: Updated street address of FSF. Reported by Xavier Bachelot. 2.0.6beta2 2011-10-26 * data/Rules-6.1.0beta.txt: - Removed LB25 entries since they are handled by hardcoded preprocessing. * debian/*: - Updated Debian packaging info by which in Debian sid packaged by Emmanuel Bouthenot. * lib/break.c: - Implemented strict LB25 rule. See also UAX#14 8.2 Example 7. - Moved linebreak_lbrule() from linebreak.c to break.c. - For compatibility, linebreak_lbrule() gives results by simplified rules LB1, LB21a & LB25 and emulates LB9, LB10. * lib/linebreak.c: - New: linebreak_get_lbrule(): Resolve AI then apply linebreak_lbrule(). 2.0.6beta1 2011-10-21 * Update UCD to 6.1.0beta at Sep 19. * lib/break.c: - New prep for LB21a about HL (as of UAX#14 rel 27; 6.1.0 draft 2). * lib/charprop.c: - Unassigned codepoints (Cn) are GCB Control, not Other (as of Unicode 6.1.0beta data). * lib/linebreak.c: - linebreak_lbrule(): treat HL as AL. * Makefile.am: - increase -version-info from 4.2.1 to 4.3.1. * src/sombok.c: - Fix SF.net bug #3400519(1): removed ESTRPIPE which is specific to Linux etc. Reported by Yusuke Muraoka (jbking@sf.net). * tests/01break.t: - Added Hebrew (he) test. * config.guess: * config.sub: - Updated. * COPYING: - Fix CPAN RT#71657: Updated street address of FSF. Reported by Xavier Bachelot. * [Note] ``Compact array'' has grown bigger than 8191 entries. Therefore the architectures with 16-bit signed pointer won't be supported anymore. 2.0.5 2011-04-23 * No new features. * New versioning scheme. Suggested by Emmanuel Bouthenot. 2011.5RC Sat Apr 16 2011 * break.c: - Bug Fix: Infinite loop occurs when prep_func generated very long (> charmax) unbreakable line. * utils.c: - linebreak_prep_URIBREAK(): loosely refers RFC3986. 2011.4 Fri Apr 01 2011 * Added Debian packaging info. * Makefile.am: - increase -version-info from 4.1.1 to 4.2.1. - added doc rules. * sombok.h.in: new macro SOMBOK_VERSION. 2011.4RC Sat Mar 26 2011 * charprop.c, break.c: Fix groundless coding denied by honest C compilers (e.g. Solaris cc). * Makefile.am: Added README.ja_JP to EXTRA_DIST. * tests/lb.sh: function definition suitable for both bash and d?ash. 2011.3 Mon Mar 21 2011 * tests/01break.t: Added tests for Vietnamese. 2011.3RC Sun Mar 13 2011 * configure.ac: unichar_t may be unsigned int or unsigned long. * include/sombok.h: now generated from include/sombok.h.in. * Bug Fix: southeastasian.c: th_wbrk() may return 0 _or_ -1 when breaking position was not found. * break.c: some optimizations. 2011.2 Sat Mar 05 2011 * charprop.c: Fix: _add_prop() broke custom map. 2011.2beta1 Sat Feb 26 2011 * charprop.c: New: linebreak_update_lbclass(), linebreak_clear_lbclass(), linebreak_update_eawidth(), linebreak_clear_eawidth(). * tests/03ns.t: added test on custom map. * tests/06context.t: added test on custom map. * Typos: GBC will be GCB. * Chg: update -version-info to 4:1:1. 2011.1 Sat Feb 19 2011 * Chg: update -version-info to 4:0:1. * Doc: some additions. 2011.1RC Fri Feb 11 2011 * break.c: New: ``preprocess'' callback feature. user_func is obsoleted. * utils.c: built-in preprocess callback linebreak_prep_URIBREAK(). * sombok.c: custom format callback using shell script. * Test: added test data for ``make check''. 2011.0 Wed Jan 26 2011 * Brush-up documentations. * ``make check'' feature (incomplete). 2011.0RC Thu Jan 20 2011 * Fix: Eliminated all (maybe, ...hopefully) memory leaks. * Chg: split part of gcstring_substr() to gcstring_replace(). 2011.0b1 Sun Jan 15 2011 * First beta of new major release. Important Changes beside Linebreak 1.x: * 1.x has memory leaks & potantial buffer overrun vulnerabilities. Switching to release 20XX is strongly recommended. * Renamed package name from Linebreak to Sombok. * Major changes on public interface: - urgent_func callback takes two arguments, not five. - sizing_func callback with extra 6th argument was obsoleted. - default sizing method simply returns number of grapheme clusters, not number of columns based on UAX #11. For compatible behavior with 1.x, use linebreak_sizing_UAX11() built-in callback function. - linebreak_break*() functions return array of lines, not single Unicode string. Additinally, elements of array are grapheme cluster strings. About change history until release 1.x see ChangeLog.REL1 file.