Fredrik Roubert [Wed, 10 Jun 2015 17:21:08 +0000 (17:21 +0000)]
ICU-11734 Bugfix: RelativeDateTimeFormatter with U_HIDE_DRAFT_API.
RelativeDateTimeFormatter has a member of type
UDateRelativeDateTimeFormatterStyle, so this type must be declared even
when setting U_HIDE_DRAFT_API, but it must then have a dummy enum value,
for empty enum declarations are not valid.
Fredrik Roubert [Wed, 10 Jun 2015 17:20:24 +0000 (17:20 +0000)]
ICU-11730 Bugfix: UNumberFormatAttributeValue with U_HIDE_INTERNAL_API.
DecimalFormat has a member of type UNumberFormatAttributeValue, so this
type must be declared even when setting U_HIDE_INTERNAL_API, but it must
then have a dummy enum value, for empty enum declarations are not valid.
Yoshito Umaoka [Fri, 5 Jun 2015 18:45:59 +0000 (18:45 +0000)]
ICU-11733 Check null for ICU DateFormatSymbols wrapped by DateFormatSymbolsICU. So DateFormatSymbols.clone() no longer cause NPE even an instance is not fully initialized.
Markus Scherer [Fri, 29 May 2015 19:33:24 +0000 (19:33 +0000)]
ICU-11601 UConverterSharedData: add isReferenceCounted, stop checking for special referenceCounter=~0 value outside mutex to make thread sanitizer happy, remove long-unused table pointer field, add & use initializer macro
Markus Scherer [Thu, 28 May 2015 23:17:23 +0000 (23:17 +0000)]
ICU-11574 near-final Unicode 8 data with new block API constants, code adjustments & fixes; svn merge --reintegrate branches/markus/uni80; plus Java port of C++ r37450 spoof data d6 update
Yoshito Umaoka [Wed, 20 May 2015 23:44:16 +0000 (23:44 +0000)]
ICU-11602 Fixed a wrong condition used for detecting keyword without value in uloc internal function for parsing keywords. TestKeywordVariants was updated to include a case hitting the code path. The test code was also updated to handle error status. A follow up ticket #11690 was created for recording some issues with the test case found in this time.
Yoshito Umaoka [Tue, 19 May 2015 21:25:21 +0000 (21:25 +0000)]
ICU-11677 Added a couple of test cases parsing abutting numeric date field patterns including quarter - Qyy and QQyy in TestQuarters(). ICU4C did not have the parsing problem reported by #11677, but I want to add some test cases matchnig the ones added in ICU4J side.
Yoshito Umaoka [Tue, 19 May 2015 21:25:08 +0000 (21:25 +0000)]
ICU-11677 Fixed numeric field detection method to return true for pattern Q up to two letters. The pattern letters constant were updated to match the corresponding C implementation (DateFormatSymbols:isNumericField(UDateFormatField,int32_t)). Uncommented TestQuarters which was commented previously for some reasons, and added the problematic abutting patterns - Qyy and QQyy in the test data.
Fredrik Roubert [Tue, 12 May 2015 18:38:39 +0000 (18:38 +0000)]
ICU-11674 Make ULocale tests work better with Android.
ULocale has been updated to work with scripts and Unicode extensions in
the Android Locale, so the tests can be updated to expect this to behave
essentially like Java 7.
Fredrik Roubert [Fri, 8 May 2015 14:43:48 +0000 (14:43 +0000)]
ICU-11674 Make ULocale work better with Android.
Android supports scripts and Unicode extensions, but not locale
categories. Instead of identifying Java version to determine
capabilities, this change introduces two booleans in JDKLocaleHelper,
one that tells whether the implementation supports locale categories,
and another that tells whether scripts and extensions are supported.
George Rhoten [Thu, 23 Apr 2015 16:13:46 +0000 (16:13 +0000)]
ICU-11564 Improve the thread safety of RBNF.
The recursion count is now a method argument instead of a data member.
Setting the DecimalFormatSymbols no longer reparses the rules, but it just sets a new version of DecimalFormatSymbols instead.
George Rhoten [Thu, 23 Apr 2015 07:44:14 +0000 (07:44 +0000)]
ICU-11564 Improve the thread safety of RBNF.
The recursion count is now a method argument instead of a data member.
Some data fields are now marked final to ensure thread safety, which caused reordering of some data initialization/parsing.
Setting the DecimalFormatSymbols no longer reparses the rules, but it just sets a new version of DecimalFormatSymbols instead.
Markus Scherer [Tue, 21 Apr 2015 23:55:18 +0000 (23:55 +0000)]
ICU-11574 fix Cherokee case folding data: add scf->self mappings for characters that do not have sfc (map to self) but have slc (ICU code: sfc falls back to slc)