Time database: Difference between revisions

From annawiki
Line 40: Line 40:
* limited support for pre-1970 data
* limited support for pre-1970 data
|-
|-
| Provide time zone acronyms that are unique within each country and at any given point in time refer to only one offset from a base time
| For any point in time, a pair of country code and time zone acronym refers to one offset to exactly only one offset from a base time
| Uses EST and CST for time stamps of zones in Australia that use DST and those that do not. That means for a given point in time during summer the offset from UTC and therefore UTC itself cannot be derived from the local time representation.
| For some points in time during DST observance in Australia, the acronyms EST and CST for time records in Australia can refer to different offsets from UTC, depending on whether a zone observes DST or not. That means for a given point in time during summer the offset from UTC and therefore UTC itself cannot be derived from the local time representation.
|-
|-
| ISO 3166-1 alpha-3 codes for countries
| ISO 3166-1 alpha-3 codes for countries

Revision as of 2013-09-20T20:44:47

Mailing list

Desired features

Time keeping database (tkdb) IANA time zone database (tzdb)
Exactly one term for one kind of object
  • Terms are used interchangeably (region = zone)
  • Terms are used ambiguously (zone for IANA zones and real world zones)
Smallest geographic unit is named "sector" Smallest geographic unit is named "zone"
The sectors cover the whole surface of the earth Some areas are not covered, e.g. the ISO 3166 country "Bouvet Island"
Each sector has exactly one sector ID. Some zones have multiple IDs via links. These links can
  • represent former IDs of the zone due to spelling change for the reference location
  • represent former IDs of the zone due to reference location change (Tel Aviv -> Jerusalem)
  • represent IDs of former zones (Vaduz links to Zurich)
A sector ID only contains alphanumeric characters, not case sensitive. (Note: further restriction likely) A zone ID contains ASCII letters and / and can contain -._
New sectors are created only by splitting and deprecating old sectors, so a user can know if an assignment of an ID to an object needs a check for correctness.
  • User can link an object to a tzid, but get's no notification if later the object is located in another zone having a different tzid
  • Different locations are linked, e.g. Vaduz points to Zurich, Tel Aviv to Jerusalem
Relations between deprecated sectors and the sectors that have been created out of them are published. No split history is explicitly published.
Record all legal time
  • no time for Bouvet Island, while Norwegian government has defined it,
  • limited support for pre-1970 data
For any point in time, a pair of country code and time zone acronym refers to one offset to exactly only one offset from a base time For some points in time during DST observance in Australia, the acronyms EST and CST for time records in Australia can refer to different offsets from UTC, depending on whether a zone observes DST or not. That means for a given point in time during summer the offset from UTC and therefore UTC itself cannot be derived from the local time representation.
ISO 3166-1 alpha-3 codes for countries ISO 3166-1 alpha-2 codes for countries. Some of the codes have been re-assigned, e.g. CS can refer to Czechoslovakia or Serbia and Montenegro.
Support for former countries at least back to 1946-01-01, the first full year where the UN existed. No support for former countries. No way to get data for ISO 3166 alpha-2 code DD (East Germany)
Publication in
  • SQL (?)
  • as xCal/iCal (?)
  • IANA specific format, as far as that format supports the features of the tkdb.
Publication in IANA specific format.