Metazone-Reform

We do not have good coverage for metazones. I've been trying to determine the set of metazones that need to be translated as the highest priority. In so doing, this involves decideing where we really need metazones, and where we don't. This is for use with the XMB translation, but it has overlap with the other coverage work we're doing.

Here are my thoughts. 
  1. The country-name-style works pretty well as a fallback, and doesn't need extra strings translated (11:00 Cuba Time)
  2. If a country has a single timezone*, use the fallback. (**We can remove existing metazones for these cases)
  3. Some countries have a small population. Skip them, since the fallback is sufficient. 
  4. Some countries, practically speaking, have a single zone with some outlying regions of small population. Skip them, since the fallback is sufficient. 
  5. This gives the following countries: AR, AU, BR, CA, ID, KZ, MN, RU, US
  6. Translate the metazones with golden zones in those countries. That gives 55 metazones.
  7. Of those:
    1. If the golden zone has daylight savings*, then translate 3 strings: genericstandard, and daylight
    2. Otherwise, translate 1 string: generic
    3. This would involve some rework of existing data.
The * is because we don't care about historic differences. I'm recommending using the period 2010 to 2015, because that is where we care most about it.

The current breakdown according to this is in the following spreadsheet: http://goo.gl/14XR9



I ran into some oddities along the way.
  1. I stumbled with the function myZone.useDaylightTime. It doesn't mean "inDaylightTime=true for some date". Nor does it mean "inDaylightTime=true at the moment". What it appears to mean is that the current rules in effect allow for daylight time, which is sort-of useless; it doesn't tell us much in practical terms.
  2. Antarctica/South_Pole is missing a metazone. We should have a release test to make sure that all the metazone data is consistent.
  3. I think we should change some of the metazones; I think we should rework as follows, to reflect the best state of affairs after 1983:
Proposal:
Alaska - delete
Hawaii_Aleutian => Hawaii
Alaska_Hawaii => Alaska

[JCE] I don't think this is correct.  The current ones ( in modern day usage ) are "Alaska" and "Hawaii_Aleutian".  The "Alaska_Hawaii" one is the old one that was in effect prior to 1983.

We don't actually need the old "Alaska", since it wasn't in use before 1983. "Hawaii-Aleutian (Standard) Time" is used, but far less than "Hawaii (Standard) 
Time".

Here is the data:

Region  Metazone   Golden_Zone
US Alaska America/Juneau
US Alaska_Hawaii America/Anchorage
US Hawaii_Aleutian Pacific/Honolulu

America/Adak: {{-∞, 1983-10-30 12:00}, Bering}

America/Adak: {{1983-10-30 12:00, ∞}, Hawaii_Aleutian}


America/Anchorage: {{-∞, 1983-10-30 11:00}, Alaska_Hawaii}

America/Anchorage: {{1983-10-30 11:00, ∞}, Alaska}


America/Juneau: {{-∞, 1983-10-30 09:00}, America_Pacific}
America/Juneau: {{1983-10-30 09:00, ∞}, Alaska}


America/Nome: {{-∞, 1983-10-30 12:00}, Bering}

America/Nome: {{1983-10-30 12:00, ∞}, Alaska}


America/Yakutat: {{-∞, 1983-10-30 10:00}, Yukon}

America/Yakutat: {{1983-10-30 10:00, ∞}, Alaska}


Pacific/Honolulu: {{-∞, 1983-10-30 11:00}, Alaska_Hawaii}

Pacific/Honolulu: {{1983-10-30 11:00, ∞}, Hawaii_Aleutian}


Pacific/Johnston: {{-∞, 1983-10-30 11:00}, Alaska_Hawaii}

Pacific/Johnston: {{1983-10-30 11:00, ∞}, Hawaii_Aleutian}


<metazone type="Alaska">

<long>

<generic>Alaska Time</generic>

<standard>Alaska Standard Time</standard>

<daylight>Alaska Daylight Time</daylight>

</long>

<commonlyUsed>true</commonlyUsed>

</metazone>

<metazone type="Alaska_Hawaii">

<long>

<generic>Alaska-Hawaii Time</generic>

<standard>Alaska-Hawaii Standard Time</standard>

<daylight>Alaska-Hawaii Daylight Time</daylight>

</long>

</metazone>


<metazone type="Hawaii_Aleutian">

<long>

<standard>Hawaii-Aleutian Standard Time</standard>

</long>

<commonlyUsed>true</commonlyUsed>

</metazone>


Comments