this PR is related to #65, and fixes #68 and #69, and partially solves #75

another alternative is to continue the work on #77

@caridy what's the gz size difference between the current and this new data for say en-US?

Hmm.. #69 (and #88) seems not to be fixed by this change, am I missing something?

$ npm install
$ $(npm bin)/grunt update-cldr-data
$ $(npm bin)/grunt cldr
$ $(npm bin)/grunt
$ node -v
> require("./index")
{ __localeSensitiveProtos: 
   { Number: { toLocaleString: [Function] },
     Date: 
      { toLocaleString: [Function],
        toLocaleDateString: [Function],
        toLocaleTimeString: [Function] } },
  default: [Circular] }
> new Intl.DateTimeFormat("ja",{year:"numeric",month:"long",day:"numeric",weekday:"long"}).format(new Date())
'2015年3月月28日(土曜日)'

Expected result as follows (Chrome's result).

> new Intl.DateTimeFormat("ja",{year:"numeric",month:"long",day:"numeric",weekday:"long"}).format(new Date())
"2015年3月28日土曜日"

@okuryu this is really not much about the short variation (yes, I have a patch in there to add few shorts variations, but this is really about the new build process so we can start making significant changes again.

To be more specific, we want to use CLDR data to support all variations, by simply using CLDR data directly instead of a computed portion of it. Here you can see the big difference from one locale to another:

https://github.com/unicode-cldr/cldr-dates-full/blob/master/main/en-001/ca-gregorian.json#L309-L349
https://github.com/unicode-cldr/cldr-dates-full/blob/master/main/ja/ca-gregorian.json#L300-L345

As you can see, they don't have the same entries, instead, we just need to update our computations, to pick up the right variation from the list available instead of a predefined, prebuild subset.

This PR seems to be complete now, although functional tests (saucelab tests are failing, I might need help with that since I don't have too much time before going on vacation).

I'm +1 for whatever we need to do to make Intl.js act like browsers do.

fixes #77

fixes #75

fixes #69

fixes #68

@juandopazo is working on the saucelab failures which is the only missing bit at this point for rc1 to be released.

We were able to test saucelabs manually, we just need to figure what to do with the travis integration that is failing today. I will merge this and figure that in a different PR.

ping @caridy it'd be awesome if u can merge this 👍 thanks!

Thanks @andyearnshaw. When can we expect it to be published?

@caridy it seems that there's missing src file src\cldr.js. With latest source code grunt build fails with this error

Running "bundle_jsnext:dest" (bundle_jsnext) task
INFO: Using a global wrapper module for "src/main.js", exporting namespace "IntlPolyfill"
Fatal error: Error writing bundle in "dist/Intl.js": Error: missing module import from ./core for path: ./cldr

Can you please release a new version/tag with this fix? Thank you.

publish it please 👍

yep a new npm version with these fixes would be great
Thks for the hard work!

1.0.0-rc-1 is in NPM, help us to test it...

works fine for me, thks a lot

Issue ember-intl/ember-intl#101 is partially fixed (works fine on Chrome and IE11, not on IE9 and IE10).

Regards,

@bcecchinato thanks for the report. Can you give us more details about the issue wiht IE9/IE10?

@caridy : When using the following format for date :

export default {
    'date-style': {
      day:  '2-digit',
      month:  '2-digit',
      year:  'numeric'
    }
};

Day and month are on one digit when lower than 10. For instance, 15/06/2015 would be written 15/6/2015. The issue was also occurring on Chrome and IE11, since the 1.0.0-rc-1 it seems to be fixed.

Regards,

can you open an issue for that so we can try to fix it. I have some suspicious about it, but I will have to dig in.

@caridy Done in issue #105 :)