<i18n dev> [8] Review request for JEP 127: Improve Locale Data Packaging and Adopt Unicode CLDR Data

[Erik Joelsson]

And I'm done.

http://cr.openjdk.java.net/~erikj/6336885/webrev.build-infra.00/ 
<http://cr.openjdk.java.net/%7Eerikj/6336885/webrev.build-infra.00/>

This is a patch on top of your latest webrev that I have verified to 
produce the same result on linux. This part will need some more eyes 
from build-infra on it too.

I also changed make/java/java/localegen.sh, readding the sort to make 
builds easier to compare. This also has the added benefit of keeping the 
build more reproducible and less dependent on the file system of the 
build host.

/Erik

On 2012-08-14 08:40, Erik Joelsson wrote:
> I finished making the changes yesterday, but didn't manage to get the 
> patch completed before I had to run. (webrev and mercurial wouldn't 
> cooperate). I will get it done asap today.
>
> /Erik
>
> On 2012-08-13 23:43, Naoto Sato wrote:
>> Since I haven't heard any more comments from Erik/Kelly, I would like 
>> to push the changeset without the new build infra patch. Erik/Kelly, 
>> can you please give us an official "GO" in terms of the build related 
>> changes?
>>
>> Aside from build changes, I have updated the changeset based on an 
>> internal review:
>>
>> http://cr.openjdk.java.net/~naoto/6336885/webrev.01/
>>
>> This includes:
>>
>> - a couple of fixes to the CLDR Converter
>> - Added fallback for any bad SPI implementations which return null 
>> for requested instances.
>>
>> Still asking for review comments.
>>
>> Naoto
>>
>> On 8/8/12 2:13 PM, Naoto Sato wrote:
>>> On 8/7/12 2:57 AM, Erik Joelsson wrote:
>>>> See inline
>>>>
>>>> On 2012-07-13 22:20, Kelly O'Hair wrote:
>>>>> Something seems strange here:
>>>>> http://cr.openjdk.java.net/~naoto/6336885/webrev.00/make/java/java/localegen.sh.sdiff.html 
>>>>>
>>>>>
>>>>>
>>>>> It's like someone was avoiding overall quotes, but using them to add
>>>>> spaces somehow...
>>>>> I sure would like to get rid of this shell logic, seems like there 
>>>>> are
>>>>> lots repeated logic that
>>>>> this script does over and over or could be done with makefile pattern
>>>>> subst's instead of exec's.
>>>> The new version isn't any worse than the old in my opinion. In
>>>> build-infra, this file is indeed replaced with make logic. After 
>>>> having
>>>> decoded both versions I'm confident in converting the changes.
>>>>> Overall, just looking at the makefiles, the build-infra team may need
>>>>> some time to fully absorb this into the
>>>>> new makefiles, some of it will be trivial, not sure all of it will 
>>>>> be.
>>>>>
>>>> I have applied the patch to a clone of build-infra and done the 
>>>> minimal
>>>> changes to keep build-infra building, which was rather trivial. The
>>>> resulting build has large differences since I haven't converted all of
>>>> it yet. There are a couple of things that will require some more work,
>>>> but not more than a day or two.
>>>>> Not sure how to proceed here, the build-infra team does need an 
>>>>> action
>>>>> item to deal with this, maybe
>>>>> before it gets integrated because I suspect the new makefiles will
>>>>> break with all the filename or
>>>>> directory changes. But I hate to hold up your integration plans.
>>>> I see these as possible options:
>>>>
>>>> 1. Let this go in, build-infra will break in jdk8 until we do our next
>>>> push and it trickles through the repos.
>>>> 2. The build-infra project provides a patch with the full conversion
>>>> that can go in together with these changes.
>>>> 3. The build-infra project provides a simple patch which just keeps 
>>>> the
>>>> build-infra-build from failing that can go in with these changes.
>>>>
>>>> The conversion needs to happen regardless of option. The changes
>>>> required are pretty isolated from remaining build-infra work. What do
>>>> you think?
>>>
>>> Either way is fine with me. If build-infra team prefers 2 or 3, please
>>> give me the patch, and I will go ahead and merge them to my changeset.
>>>
>>> Naoto
>>>
>>>>
>>>> Other than that, I have no objections to the review.
>>>>
>>>> /Erik
>>>>
>>>>> I'd like to get some advice from Erik on this before saying anything
>>>>> further.
>>>>>
>>>>> -kto
>>>>>
>>>>> On Jul 10, 2012, at 1:42 PM, Naoto Sato wrote:
>>>>>
>>>>>> Hello,
>>>>>>
>>>>>> Please review the JDK8 changes for JEP 127: Improve Locale Data
>>>>>> Packaging and Adopt Unicode CLDR Data
>>>>>> (http://openjdk.java.net/jeps/127). The webrev is located at:
>>>>>>
>>>>>> http://cr.openjdk.java.net/~naoto/6336885/webrev.00/
>>>>>>
>>>>>> The main bug id for this enhancement is:
>>>>>>
>>>>>> 6336885: RFE: Locale Data Deployment Enhancements
>>>>>>
>>>>>> Along with this, the following bugs/enhancements are also 
>>>>>> implemented
>>>>>> in this change:
>>>>>>
>>>>>> 4609153 Provide locale data for Indic locales
>>>>>> 5104387 Support for gl_ES locale (galician language)
>>>>>> 6337471 desktop/system locale preferences support
>>>>>> 7056139 (cal) SPI support for locale-dependent Calendar parameters
>>>>>> 7058206 Provide CalendarData SPI for week params and display field
>>>>>> value names
>>>>>> 7073852 Support multiple scripts for digits and decimal symbols per
>>>>>> locale
>>>>>> 7079560 [Fmt-Da] Context dependent month names support in
>>>>>> SimpleDateFormat
>>>>>> 7171324 getAvailableLocales() of locale sensitive services should
>>>>>> return the actual availability of locales
>>>>>> 7151414 (cal) Support calendar type identification
>>>>>> 7168528 LocaleServiceProvider needs to be aware of Locale extensions
>>>>>> 7171372 (cal) locale's default Calendar should be created if unknown
>>>>>> calendar is specified
>>>>>>
>>>>>> Please note that packaging changes that relate to Jigsaw module
>>>>>> system aren't included in this changeset.
>>>>>>
>>>>>> Naoto Sato and Masayoshi Okutsu
>>>
>>