Thanks Masayoshi for the review and comment.
-michael
On 14年01月08日 01:37 上午, Masayoshi Okutsu wrote:
Hi Michael,
This fix looks good to me.
if the users hardcode the names in their program or tests,
I don't think names hardcoded by apps should be concerned. My concern
is names produced using the public API. For example, two apps (client
and server) running on different Java versions may not be able to
perform consistent display name-based operations. Name changes should
be minimized for fixing wrong names IMHO.
Thanks,
Masayoshi
On 1/8/2014 1:02 PM, Michael Fang wrote:
Hi Masayoshi,
Since the changes to the LocaleNames translation might have slight
chance of backward compatibility risk (if the users hardcode the
names in their program or tests, both not recommended), I have
decided to re-do the work and only apply missing names to existing
files. The webrev is much cleaner now:
http://cr.openjdk.java.net/~mfang/8029239/webrev.jdk.04/
http://cr.openjdk.java.net/~mfang/8029239/webrev.jdk.04.utf8/
If there are missing script names, they need to be added to English
LocaleNames.properties file first, then I can coordinate the
translation. This will happen after jdk8.
thanks,
-michael
On 13年12月04日 11:23 下午, Masayoshi Okutsu wrote:
Thanks for making the \uxxxx changes. It's much easier to review.
I checked LocaleNames_ja.properties with JIS X 0412 (Japanese
version of ISO 639) and JIS X 0304 (Japanese version of ISO 3166).
Many changes corrected wrong names, but there are still questionable
ones, and some are inconsistent with JIS X 0412/0304, such as 中華人
民共和 国 to 中国 for CN because 中国 is too ambiguous [1]. My
preference is to follow the JIS standards (where applicable).
Also I noticed some new script names are missing, such as Jurc for
Jurchen. Should this problem be out of the scope with the 8029239 fix?
Thanks,
Masayoshi
[1]
http://ja.wikipedia.org/wiki/%E4%B8%AD%E5%9B%BD_%28%E6%9B%96%E6%98%A7%E3%81%95%E5%9B%9E%E9%81%BF%29
On 12/5/2013 8:02 AM, Michael Fang wrote:
Thanks Masayoshi for the feedback.
I looked at the existing LocaleNames_xx.properties files and it
contains lower case \uxxxx notation for most of sections of the
files, but newer sections as ISO 639.2 language code and ISO 15924
script code were in upper case since they were added by the current
translation team.
I converted the files to all lower case \uxxxx notation and the
webrev didn't help (worse with even more deltas):
http://cr.openjdk.java.net/~mfang/8029239/webrev.jdk.02/
To make it easier to review, I have created updated version of
webrev that ignores cases with "diff -i" option.
http://cr.openjdk.java.net/~mfang/8029239/webrev_ignore_case.ksh
But I kept the upper case \uxxxx notation because that's what I
will be receiving from translation team going forward. In the
future, the diff should be more manageable.
The resulting webrev:
http://cr.openjdk.java.net/~mfang/8029239/webrev.jdk.03/
http://cr.openjdk.java.net/~mfang/8029239/webrev.jdk.03.utf8/
I need to check with translation team about the Japanese
translation issue for "afa".
I have added bug# to test/sun/text/resources/LocaleDataTest.java
thanks,
-michael
On 13年12月04日 01:17 上午, Masayoshi Okutsu wrote:
I still don't like the unnecessary case changes for the \uxxxx
notation, which makes code review very difficult (even UTF-8 ones).
I noticed the following change uses '=' (the second one) as a word
delimiter. This looks strange to me.
src/share/classes/sun/util/resources/ja/LocaleNames_ja.properties:
-afa=\u30A2\u30D5\u30AC\u30CB\u30FC (1927-2002)
+afa=\u30A2\u30D5\u30ED=\u30A2\u30B8\u30A2\u8A9E\u65CF
Should the bug# be added to
test/sun/text/resources/LocaleDataTest.java when LocaleData has
been modified?
Thanks,
Masayoshi
On 12/4/2013 1:02 PM, Michael Fang wrote:
I have updated the webrev
http://cr.openjdk.java.net/~mfang/8029239/webrev.jdk.01/
2 failed regression tests have been addressed:
----------System.out:(7/192)*----------
Mismatch in LocaleNames/pt/wa:
file = "val\\\\u00e3o"
jvm = "val\\u00e3o"
Mismatch in LocaleNames/pt/FM:
file = "Micron\\\\u00e9sia"
jvm = "Micron\\u00e9sia"
Test failed. 2 errors.
thanks,
-michael
On 13年12月03日 05:10 下午, Michael Fang wrote:
Hi,
Please help to review the changes for the following CR:
https://bugs.openjdk.java.net/browse/JDK-8029239
The webrev is available here:
http://cr.openjdk.java.net/~mfang/8029239/
The changes included the following:
Portuguese files were updated with resources generated from CLDR
2.1. LocaleNames_pt_BR.properties have been removed as a result.
Regression tests were performed and failed tests were updated to
be compatible with the changes to CLDR 2.1.
There are also translation capitalization changes. I did some
research about capitalization rules and found that for Spanish,
French, Italian, Portuguese, and Swedish, language names are not
capitalized, but country names are and these are consistent with
CLDR as well.
References about capitalization
http://meta.wikimedia.org/wiki/Capitalization_of_Wiktionary_pages#Languages_without_capitalization:
es, fr, it, sv, pt, etc
Spanish:
http://www.howto.gov/web-content/multilingual/spanish-guide/capitalization
French:
http://french.about.com/library/writing/bl-capitalization.htm
Italian: http://www.translationdirectory.com/article714.htm
Swedish: http://en.wikibooks.org/wiki/Swedish/Nationalities
There are also translation changes performed by translators. I
reviewed each change against CLDR. I only accepted the changes
that's consistent with CLDR since I don't know the languages and
could only use CLDR as a reference.
I did not add new regression tests for the translation changes
since there are no formatting changes, just translation changes.
We have a keyword noreg-l10n that covers the exception for
requiring regression tests for translation changes.
thanks,
-michael
