From OpenStreetMap Wiki
Jump to: navigation, search

Timezone boundaries not observable ?

They are obviously observable on REALLY MANY (tens of millions!) local clocks.

Even if there no visible border, this is the case of almost all administrative boundaries (including most parts of country boundaries) that are only visible in a few very local places.

But I agree that we don't really need these as boundaries, but possibly only as properties of countries/states/territories, i.e. a single tag on them (ideally the identifier from the wellknown Olson's database, as seen in the CLDR or IANA timezones databases. (Most countries or relevant subdivisions already have such tags). See Key:timezone instead to tag relevant countries/territories or subdivisions.

A good reason for that is that timezone boundaries would be complex and giant relations, very difficult to edit and in fact difficult to use. It is much simpler to add a single tag on the maintained administrative subdivisions (which are already complex and already large enough for many countries). The other reason is that they also change regularly: most softwares or OSes are updating their timezone databases per country/region.

Also locally some countries (or local communities) do not obbserve the timezone but the local solar clock, as it is observed locally: official timezones are only defined for a specific place in the country (generally the capital city), even if local time is solar, with the primary intent to synchronize and secure transportation or for legal reasons (court orders, application of laws, commercial contracts), or apply some work time regulation. So local clocks display this legal time. Solar time tends now to be used more and more culturally (for religious purpose mainly) and people don't like having two local clocks. These legal times are simple offset adjustments from UTC (multiples of hours, sometimes half-hours or quarters of hours).

Verdy_p (talk) 10:19, 30 January 2017 (UTC)

One note: timezone boundaries have been currently defined in large countries: Russia, Brasil, Mexico, some parts of Canada and USA. They do not match exactly with adminsitrative boundaries (notably in Russia and Canada, but also in Western Brasil, or Greenland, part of Denmark).
Map with Overpass Turbo. (warning: about 70MB of data, takes long to render and requires an efficient browser and lot of memory, but works in Chrome even in 32-bit mode).
Timezones that are set for Spain and Portugal are not needed at all (they match the country subdivision borders exactly). Same thing about Argentina (they match regional subdivisions).
So in my opinion timezone boundaries should not be completely excluded, unless they match an existing admin boundary: I think that they are good for Russia, Canada, Greenland and even for USA and Brazil. For Mexico (fully covered by them) they match the Mexican states exactly (with one exception, possibly a group of subdivisions in one state) and can be removed as unnecessary duplicates (to be replaced by timezone=* tags).
I then oppose deleting this page. But instead of keeping as it was, just explain better when they are not necessary. — Verdy_p (talk) 11:17, 30 January 2017 (UTC)
Timezones also stretches through seas and oceans where no administrative boundaries present. --Keder (talk) 18:54, 1 February 2017 (UTC)
Timezones on seas are not definitive, notably in international waters ! The applicable local timezone on ships will change according to decision by navigators. For radio-communications, they use UTC, or use the timezone of the nearer harbour or controling authority. But most of the time the timezone will change during the night, between peak periods of work. Some ships will not even change their local timezon on board when coming to an harbour for a short period of time, or if they don't reach the coast and continue navigating, or will skip one timezone to go directly to the next one or long trips if they don't reach any harbour in the middle of the course.
There's in fact no local timezone applicable on a map boundary for international waters: using UTC-based timezones based purely on longitude is only an arbitrary approximation (in addition DST may or may not be applied: it will apply if it applies both at the departure and the arrival point, but there's no international agreement if this is not the case: captains on each ship decides its local working schedule).
For this reson we don't need timezones at all on international waters, but only in territorial waters (possibly extended to the EEZ where it applies, and where there's no conflict of juridiction between EEZ claims). For air navigation, timezones applies are those from controling authorities, but most of them use UTC only to avoid confusions, and local time on board will only change on arrival in airports just before starting the landing procedures). — Verdy_p (talk) 17:00, 2 February 2017 (UTC)
Note: I've reverted the mere blanking of this tag description page made by Fredrick Ramm (but kept his deletion request at top). In my opinion, such deletion request by blanking contents is highly contestable when there's evidence that this tag is used, and relevant. Alternatives requre further developments. Notably these boundaries are used to resolve other tags based on local time, such as opening_hours=*, and there's no easy replacement without tagging lots of administrative boundaries (not just countries but smaller regions or complex sets of regions, as well as non administrative areas such as parts of oceans) with their relevant timezone, and in some cases imezones are overlapping (and coexist on the same places, so these places cannot be tagged with a single timezone: in such cases, local time values will need to specify the relevant timezone, or some arbitration process will need to determine it depending on usage or standards in use by application, such as air or railway transportation, legal/accounting, emergencies, prayers..., each one with its own applicable boundaries).
In frequent cases, the admin boundaries do not even exist in OSM (or are extremely fuzzy/unstable), while timezone boundaries are much more stable (but they change as well, not necessarily in sync!).
In summary timezones are independant boundaries (still most frequently international), and very frequently overlapping almost everywhere: each one has its own definition of their applicable area.
A deletion will require further discussions notably for migrating aplciations depending on them.
This does not mean that the timezone tag cannot be set on relevant country/territory admin boundaries (in which case, these tags will override what timezone boundaries indicate for the legal local timezone). But trimezones are not just for legal time. — Verdy_p (talk) 00:27, 3 February 2017 (UTC)

Appears I may have been over-eager here to request deletion. I'm very much against mapping time zones as boundaries in OSM, and sadly meanwhile the usual drive for completeness seems to have kicked in and some users are adding time zone boundaries even in places where it would be sufficient to tag a handful of states with a time zone. But I must admit I thought that time zones would always run along admin boundaries (smaller ones maybe, but some admin boundary); in user Shinigami claims that this is not the case. If he is right then there might indeed be some use for a boundary=timezone tag. I wonder about the legal/copyright situation though and will quizz user Shingami on this in the further changeset discussion. I think the whole matter of time zone mapping should be brought up on the tagging list. --Frederik Ramm (talk) 10:46, 10 February 2017 (UTC)

Discussion: --Frederik Ramm (talk) 06:45, 7 March 2017 (UTC)