On Mon, 18 Feb 2019 at 12:23, Guy Harris <guy@alum.mit.edu> wrote:

If so, then is the reason that we have

Link Asia/Bangkok Asia/Phnom_Penh # Cambodia
Link Asia/Bangkok Asia/Vientiane # Laos

but don't have, for example,

Link Asia/Bangkok Asia/Hanoi # (North) Vietnam

that Asia/Phnom_Penh and Asia/Vientiane existed as separate entries before being combined with Asia/Bangkok, so that we need to keep their entire around for backwards compatibility, but there had never been an Asia/Hanoi?

It looks like that's indeed what happened. Thanks for jogging my memory. So I guess it wasn't an omission so much as an awkward and poorly-documented decision:

Asia/Phnom_Penh and Asia/Vientiane (which had existed prior) were combined with Asia/Bangkok on 2014-10-04 in https://github.com/eggert/tz/commit/6f7ddecfb036e13a1132181cfaa8ab3e070fc9e8
Asia/Hanoi (which had NOT existed prior) was added to backzone on 2014-10-11 in https://github.com/eggert/tz/commit/bcb616ca405583de10e6e8c85951720ef066877a

If we had known about the differences in Asia/Hanoi at least one release prior to this merging (and I recall all of this came up as part of the same discussion), then Asia/Hanoi would have needed to become a separate zone at that time. But because we first took the opportunity to coalesce the other zones, Asia/Hanoi's history could be coalesced as well, which is why it landed only in 'backzone' and not in 'asia'.

I do think this boils down to a judgement call, though. There certainly could be an appearance of bias (though none was intended, it simply wasn't necessary to include Asia/Hanoi for backwards compatibility as it hadn't yet existed). Given that and the other complexities here, I would personally be inclined to move the Asia/Hanoi into the mainstream distribution, but I can absolutely see how others involved in the maintenance here may disagree. At the very least, better documentation in the 'asia' file is a pretty clear must.

--
Tim Parenti