<i18n dev>  RFR: 8222756: Plural support in CompactNumberFormat
Roger.Riggs at oracle.com
Thu Dec 5 20:13:57 UTC 2019
Thanks for the updates.
On 12/5/19 1:49 PM, naoto.sato at oracle.com wrote:
> Thanks, Joe and Roger, for the reviews. Here is the updated webrev:
> These are the changes since v.01:
> - Reflected the CSR changes (thank you, JoeD, for the quick
> turnaround!), and misc typos in the spec.
> - Added length limitation to the "pluralRules" argument in the new
> constructor. Throws IllegalArgumentException if the input is too long
> (>2,048 chars).
> - Added validation to plural rules, and throws
> IllegalArgumentException if the given rules' syntax is incorrect.
> - Consolidated the same pattern to get the affixes into a utility
> method (getAffix())
> - Directly find NaN and Infinity in the number string, parse the
> number with regex otherwise.
> TestEquality.java: Tidied the test array up.
> CLDRConverter.java: Added String::trim so that the generated
> PluralRules.java will not have extra spaces.
> TestPlurals.java: Added test cases for invalid cases, i.e., invalid
> syntax and length limit exceeding rules in the constructor.
> Not addressed:
> - ResourceBundleGenerator.java exists in the webrev as it does have a
> modification (only seen in "patch" link, as the mod is only the
> - Using "==" for double values: As Roger mentioned, it is in fact
> comparing integers (the only reason to use double is to deal with NaN
> and Infinity)
> - Possible performance improvement: Could be addressed later if it is
> regarded as an issue. But since the effective plural rules are
> simple/short enough, I would not expect it as a problem.
> On 11/26/19 1:35 PM, naoto.sato at oracle.com wrote:
>> I modified CompactNumberFormat.java to simplify the syntax parsing:
>> Please review this webrev instead.
>> On 11/25/19 1:16 PM, naoto.sato at oracle.com wrote:
>>> CompactNumberFormat has been added since JDK 12 to support compact
>>> number formatting, such as 10_000 being formatted as "10K."  It
>>> works fine for English, however, not for other languages that take
>>> plural forms in formatted number prefixes/suffixes. In order to fix
>>> this, I filed the following CSR to extend the current
>>> CompactNumberFormat spec:
>>> It basically accommodates the plural prefix/suffix forms into the
>>> existing compact patterns array, so that the existing compact number
>>> format works compatibly. The plural rules are solely based on the
>>> CLDR's plural language rules 
>>> Here is the implementation of the CSR:
>>> Please review the CSR as well as its implementation.
>>>  https://bugs.openjdk.java.net/browse/JDK-8177552
More information about the i18n-dev