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* Withdrawn.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GCB Joiner and After_Joiner and LBC ZJ are withdrawn by UC.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.