brkitr/ko.txt is created to use line_*_.cj.txt for both
lw=phrase and lw != phrase cases for Korean. This is the simplest
way to fix ICU-22119 taking advantage of the fact that ICU
does not have a Korean dictionary so we don't have to worry about
adding the list of Korean particles to keep them attached to the
preceeding word.
The downside is that it only works when the locale is ko or ja while
it should work in any locale. Another is it makes ICU deviate from
CSS3 by using the same CJ (conditonal Japanese) rules for Korean as
well. However, CSS3 spec is wrong on that point and should be changed.
See https://unicode-org.atlassian.net/browse/CLDR-4931 .
In class RuleBasedBreakIterator, refactor how object initialization is handled
by the various constructors, taking advantage of C++11's ability to directly
initialize data members in the class declaration.
This will simplify ongoing maintenance of the code by eliminating the need
to keep initialization lists synchronized with the class data members.
This is being done now in preparation for additional changes to fix problems
with the handling of memory allocation failures.
In the C++ break iterator code, change all use of NULL to nullptr.
This is in preparation for follow-on PRs to improve out-of-memory error handling
in Break Iterators, keeping use of nullptr consistent between old and new
or updated code.
- rewrite to semi modular code
- run the sections first, then get counts
- section counts in ToC, skip if empty
- many link improvements: linkify bugs and commits rather than separate lines
- new option, --commit-metadata= with path to metadata file
- new option, --fix-version=41 (used for SKIP sections)
- scaffolding for 'bad commits' list
- new module CommitMetadata with unit tests
- sample file format TEST_COMMIT_METADATA.md
- such commits are skipped
- refactored the commit skipping part (formerly used for cherry pick skips)
- add a report section for skipped commits
- add a cache for JIRA queries (for dev use): --cache-for-dev "/tmp/cldr-commit-cache" - optional
- add an 'excluded commits' section at the bottom
- make sure commit metadata is used to update ticket IDs and messages.