Olson, Arthur David (NIH/NCI) [E] <olsona <at> dc37a.nci.nih.gov> writes:

I see at... http://www.usembassy.gov ...that the last likely holdout is prepared to refer to Vietnam's most populous city as "Ho Chi Minh City."

[Putting my introduction bellow since this posting system complains about top posting when it really isn't...] Hi there, I'm answering to this from Debian/Ubuntu bug reports links to your site. I'm sorry to have to tell this since it will probably be read as an agressive answer, but... Could we please stop thinking from the U.S.A. point of view only? Please see this comment from Vietnamese people: https://bugs.launchpad.net/ubuntu/+source/tzdata/+bug/121540/comments/10 And this one from myself (living in Vietnam since 1999): https://bugs.launchpad.net/ubuntu/+source/tzdata/+bug/121540/comments/15

We'd normally use "Ho Chi Minh City" (English version) rather than "Thành phố Hồ Chí Minh" (native version) as the (basis for the) zone name.

I agree on this: English is the source language. Translations are done when needed through other well defined mecanisms.

We'd normally try to limit the name to at most 14 characters.

Is there a specific reason for this? Or any official rule (POSIX?) ? Because I can see other zone names as long as 30 caracters. Eg: America/North_Dakota/New_Salem

Ho_Chi_Minh_City weighs in at 16. Ho_Chi_Minh, at eleven characters, would be within the normal limit. (Using just the name of the person after whom the city is named is something to which I'm accustomed, living as I do just a mile from Washington.)

Ok. Thanks for considering this change, but... The real change to do is to move to "Hanoi" instead! The "most populous" rule is totally arbitrary and a moving target. As Mr Vũ Đỗ Quỳnh said, Hanoi will soon become the new result from this rule! Anyway, I propose a better rule specifically for simple case like a country being in only one timezone: just use the name of the Capital City. Best regards, J.C.

Ken Pizzini <tz_@...> writes:

The constraint that is actually in play is that POSIX specifies that the length of a _portable_ path component is limited to 14 characters (bytes, really).

Thanks for this clarification. J.C.

Dave Rolsky <autarch <at> urth.org> writes:

The "most populous" rule is definitely not arbitrary.

Well... Agreed. I wrote a bit too quickly on this one... I understand your points and that you won't change the rules just for one country. So please read my answers below as some constructive criticism, not as a request to change the rules.

First of all, in the cases of larger countries with multiple zones, you can't pick the capital (the US is a good example, as are many other "wide" countries).

Agreed. But "wide" countries are not the most frequent case though... So we need some rule for them, but why should it be the same for every other "more common width" countries?

Also, one of the stated goals of the database is to "be robust in the presence of political changes". I'd assume that the most populous rule falls out from that, because capitols probably change more often than populations move.

Still this rule doesn't catch administrative changes like moving city's boundaries, which can change the "most populous" result without any people move. That should be the case for Hanoi soon. Will you change the zone name accordingly then? But I agree it's certainly more stable than using the capitol, in case of political instability, which is sadly still common...

And arbitrary or not, it's the rule, and there's no good reason _not_ to stick with it now that it's been in place for so long. Special-casing Vietnam would A) cause the maintainers a maintenance hassle;

For A) it's not our goal, of course. And as a programmer sometime, I can understand this point very well! ;-)

and B) invite everyone else to come complain that _their_ capital isn't used for the name, but Vietnam's is.

As for B), if everyone else went to complain for using their capitol name, then I would call this a real reason to make it being a rule! ;-) But let's come back to our specific problem, keeping the existing rules. It seems that we can use some kind of "links" between zone names, isn't it? So I would propose this: - rename "Saigon" as "Ho_Chi_Minh" (as already proposed, thanks) ; - make a link from "Hanoi" to "Ho_Chi_Minh". It would solve three problems at once: - the "most populous" city name change since 1975 ; - the request for "Hanoi" being usable as a zone name ; - the "soon to come" change in the "most populous" result. Regards, J.C.

Robert Elz <kre <at> munnari.OZ.AU> writes:

[...] The big advantage to the "largest population centre" is that it applies easily, and without much in the way of arguments, in all zones.

Thanks for these examples.

What might be possible would be to support "political" links, as well as backwards compatability links, and perhaps have a rule that if there is no zone named for the capital of a country, a political link for it be created (linked to whatever is the "correct" timezone according to our rules).

I strongly second this!! If it's doable without too much hassle, please do!!

Robert Elz <kre@munnari.oz.au> wrote on Fri, 21 Mar 2008 at 11:56:51 +0700 in <6916.1206075411@epsilon.noi.kre.to>:

What might be possible would be to support "political" links, as well as backwards compatability links, and perhaps have a rule that if there is no zone named for the capital of a country, a political link for it be created (linked to whatever is the "correct" timezone according to our rules). But I think we'd need to be firm about that only applying to countries, and not states/provinces/... within countries.

If we were to have political links for countries, why make them based on cities at all? We could simply have Asia/Vietnam. --jhawk@mit.edu John Hawkinson

Robert Elz schrieb:

| If we were to have political links for countries, why make them | based on cities at all? We could simply have Asia/Vietnam.

That's more or less how the timezone names started out. It changed because that doesn't work in general. Sure, it would be OK for Vietnam, but what would America/USA mean? (Or America/Canada ? etc...)

I don't think the new system really works better. It moves the problems from the database management to the people using it. How do you know you're supposed to select America/Indiana/Vincennes if you live in America/USA/Indiana/Martin_County and not America/Indiana/ Indianapolis? What happens if, e.g. Spain decides to move part of the country from CET (+01:00) to WET (+00:00)? Europe/Madrid would suddenly become wrong for all of the users without warning. If Europe/Spain was replaced with Europe/Spain/*, users could be prompted to select the correct zone. Claus

Date: Fri, 21 Mar 2008 18:49:02 +0100 From: =?ISO-8859-1?Q?Claus_F=E4rber?= <GMANE@faerber.muc.de> Message-ID: <fs0sem$6ar$1@ger.gmane.org> First, I am not going to bother to reply to your previous message, we've been through all of that before - for anyone who wants answers, look back through archives of the tz list, and they're all there. | I don't think the new system really works better. It moves the problems | from the database management to the people using it. Yes, to some extent it does. But who do you think they are? This project is collecting and keeping data on the world's timezones, as accurately as we can, and making that, and some software that interprets it available, to anyone, for free. Incorporating that into the various systems is someone else's job. Providing an end-user interface is someone else's job. Some of that is well done, some of it isn't, none of it is relevant here. | How do you know you're supposed to select America/Indiana/Vincennes if | you live in America/USA/Indiana/Martin_County and not America/Indiana/ | Indianapolis? I have no idea, I have never lived there. But that's a different issue, that's "how do I know what my timezone is", which isn't something we can answer. Nor is it our job to attempt to explain the political absurdities that lead to that mess in the first place. | What happens if, e.g. Spain decides to move part of the country from CET | (+01:00) to WET (+00:00)? Europe/Madrid would suddenly become wrong for | all of the users without warning. Yes - for all of the users in the zone that was not the same as Madrid. But with a simple question "what's the largest city/town that always has their clock set the same as yours", and for people who can actually answer the question, that will be easy. I don't know enough Spanish geography to pick the answer, but I suspect that the affected people would. That is, once they know they have to upgrade at all - that's always going to be a problem with anything that has to react to anything that gets changed arbitrarily and for reasons that cannot be predicted very far in advance (like just about any political decision.) kre

Robert Elz schrieb:

| Anyway, I propose a better rule specifically for simple case like a | country being in only one timezone: just use the name of the | Capital City. At first glance, that sounds like it would be reasonable, but in practice, it doesn't work.

What about using the "most important" city? Importance would be determined by the following factors: * number of inhabitants (most important factor) * political status (capitals get a bonus) * historical usage (cities already in use get a bonus, too) It would also explain anomalies like Europe/Berlin (should be Europe/Rhein-Rhur) and Europe/Rome (should be Europe/Milan). If you argue that these are not anomalies, then for the same arguments, Europe/London is (should be Europe/Birmingham).

Currently India has just Asia/Calcutta (which may be Asia/Kolkata after the next update), but by applying the "single zone" rule, it would have to be Asia/Delhi instead (maybe Asia/New_Delhi - that doesn't need to concern us right now).

Which is a good thing if Indians look for Asia/Delhi first because that's their national capital.

Then Asia/Calcutta (or Kolkata) would return for the eastern half, but the western would probably be Asia/Mumbai (and Asia/Delhi would vanish - or turn into a compatability link).

Well, the "most important" city in the western half would still be Asia/Delhi.

The big advantage to the "largest population centre" is that it applies easily, and without much in the way of arguments, in all zones.

Why are there anomalies like Europe/Rome and Europe/Berlin, then? Claus