Should there be a link between Asia/Hanoi and Asia/Ho_Chi_Minh in TZDB? Background of my question is an issue reported on the ThreetenBP-issue-tracker (see: https://github.com/ThreeTen/threetenbp/issues/57) where someone has observed a crash on its Android mobile phone using Asia/Hanoi as system tz (resided in OS) while the extra tz-distribution of an external library (ThreetenBP) does not contain that identifier (probably because it is in the backzone-file of TZDB). With best regards Meno
Meno Hochschild wrote:
Should there be a link between Asia/Hanoi and Asia/Ho_Chi_Minh in TZDB? ... (see: https://github.com/ThreeTen/threetenbp/issues/57) Since the same problem was reported for Asia/Yekaterinburg, I don't see why it matters whether Asia/Hanoi is in backzone or in some other file. Anyway, we're not likely to create unnecessary names merely because of improperly configured or buggy software; otherwise we'd need entries for every misspelling of every city on the planet. Asia/Hanoi is in backzone because it is beyond the scope of the tz database: Hanoi's time history since 1970 is identical to that of Asia/Bangkok, the entry that zone1970.tab uses for North Vietnam.
Thank you very much for your answer. I understand the concern to minimize the count of identifiers where possible. However, now the question arises if the rules of outdated Asia/Hanoi will be updated, too, if for example the rules of Asia/Ho_Chi_Minh might change in the future. Will the rules of backzone-identifiers be adjusted to future changes or not? I would take an answer to this question into consideration to decide if I include the backzone-file into my future tz-distributions or not.
Meno Hochschild wrote:
Should there be a link between Asia/Hanoi and Asia/Ho_Chi_Minh in TZDB? ... (see: https://github.com/ThreeTen/threetenbp/issues/57) Since the same problem was reported for Asia/Yekaterinburg, I don't see why it matters whether Asia/Hanoi is in backzone or in some other file. Anyway, we're not likely to create unnecessary names merely because of improperly configured or buggy software; otherwise we'd need entries for every misspelling of every city on the planet. Asia/Hanoi is in backzone because it is beyond the scope of the tz database: Hanoi's time history since 1970 is identical to that of Asia/Bangkok, the entry that zone1970.tab uses for North Vietnam.
Meno Hochschild wrote:
now the question arises if the rules of outdated Asia/Hanoi will be updated, too, if for example the rules of Asia/Ho_Chi_Minh might change in the future. Will the rules of backzone-identifiers be adjusted to future changes or not?
I can't promise to maintain the contents of 'backzone' with the same attention to detail as the main database. 'backzone' is outside the scope of the database. Although some of 'backzone' is quite good, other parts are dicey, and I have little doubt that it contains errors that wouldn't be tolerated in the main database. So I suggest not relying on it quite as much.
participants (2)
-
Meno Hochschild -
Paul Eggert