zone and offset designators

[LIVINGSTON Alex]

Original version sent 2006 December 09 23:22:04 GMT+11:00, but did not 
make it to list. I have modified it (including the subject) slightly 
since then.

I am not in favour of alphabetic abbreviations as designators of time 
zones or time offsets.

First, their role is not well defined. Do they apply to

A: a geographic region and its entire time-offset history and future,

B: a particular single offset from UTC/GMT, or

C: a particular set of two or more offsets from UTC/GMT and "rules" 
describing when changes from one of these offsets to another occur?

Second, there seems to be no requirement for them to be globally or 
historically unique.

I would support designators of the following kinds:

O-1: a numeric specification of a single offset from UTC/GMT (e.g. 
"-04:00")

Z-1: a specification applying to a point on the earth's surface and 
indicating the entire history (and expected future history) of the time 
kept at that point by nearby human beings, as in the tz database (as I 
understand it)

Actually, I might also consent to:

O-1A: a numeric specification of two (or more) offsets from UTC/GMT 
that apply periodically (e.g. "+10:00/+11:00"), though preferably only 
if such a specifier is unique in the context it is used in 
(+10:00/+11:00 applies both to Tasmania and mainland southeast 
Australia, but the two regions change from +10:00 to +11:00 at 
different times at present, so using that label for a 
time-specification offset as things currently stand would not be ideal 
unless it were clear from the context which changeover dates and times 
applied) and

Z-1A: a specification applying to a well-defined, invariant region of 
the earth's surface and indicating the entire history (and expected 
future history) of the time kept by human beings occupying that region

In the headings I've used to name these types of designator, the "O" 
indicates applicability to offsets and "Z" indicates primary 
applicability to zones. However, a Z (zone) designator could also be 
applied to an offset, in which case it means the offset that applies in 
that zone (which in the case of Z-1 is a point, not a region).

Type O-1 is by far the easiest to work with, and perfectly adequate for 
many single-instant specifications. Only if (1) a period of time 
(including possibly a single date) or a repeating scheduled event is to 
be specified for a particular geographic location or (2) the offset 
that will apply is not yet known with adequate certainty, as it is tied 
to yet-to-be-finalised conventions for a particular place, might a 
single offset not be up to the task. (Have I missed any?)

If it is judged that a zone designator for a time specification cannot 
be avoided, we have the problem of what time is applicable for the 
period (an hour in the majority of cases) after clocks are retarded, 
when times that have already passed in the zone are repeated. My only 
suggestion of what to do about this is to establish a convention that, 
in the absence of indication to the contrary, the second occurrence of 
the time specified should be assumed. There is also, of course, the 
related problem of what to do if a time is specified that never occurs 
in a zone, because it is skipped over when clocks are advanced. What 
else could be done besides taking such a specification to mean the 
instant of the changeover?

If some apparent standard or official abbreviated designator is not 
clear, a lot of effort seems to be put into crafting a "suitable" one 
for the field labelled (for reasons that aren't all that clear to me) 
"FORMAT" in the tzdata files. I get the impression that people derive a 
sense of security from the authority and credibility supposedly lent to 
time specifications by such labels. Only it is a false sense of 
security, I would contend. These abbreviations seem to be seized upon 
by implementors of time-management software as though "official" and 
indispensable and to be used wherever plausible. I would like to see 
such enthusiasm diverted to the use, instead, of far less ambiguous and 
far more informative and helpful designators (like type O-1 
especially).

I might also add my vehement objection to using numbers derived from 
time offsets to label zones when the offset indicated by the number 
does not always apply. An example of this is "EST5EDT". Using offsets 
to label zones seems from my experience to be almost standard practice 
in time-related software, but it is grossly misleading when, as is very 
often the case, the indicated offset simply does not apply for part, 
indeed sometimes the majority, of the year. If numeric designators of 
such zones _must_ be used, then _both_ (or all, should there ever be 
more than two) offsets should be indicated (with equal emphasis).

I would, as well, completely avoid the use of the term "standard" to 
distinguish one of two (or more) offsets that apply in a given zone. It 
may sometimes be a designation given by a relevant authority, but it is 
by no means always used, and it is a very poor choice for a generic 
designation of the less advanced of two offsets that apply at a 
particular location, since it is far from clear that a standard exists 
from which the offset that applies could be unambiguously inferred. I 
think any attempt at pinning down a particular offset as "standard" for 
a place or region based on longitude should be abandoned. Exceptions - 
cases of times being kept that deviate considerably from what one would 
expect (if it is clear at all) given the longitude - are not rare but 
abundant. Let me list some: the Aleutians, Alaska, Argentina, the 
Azores, Iceland, Madeira, the Canary Islands, Spain and France, 
peninsula Malaysia and Singapore, central Australia, parts of 
far-eastern Russia, the Chatham Islands, and, of course, though not 
true of the zone as a whole, western China. The term "standard" is 
applicable to whatever offset is being observed: it is the standard set 
by those (in authority) observing it (and their subjects). I suggest 
the terms "basis" and "advanced" instead as generic distinguishers of 
the two offsets that are periodically observed at any particular 
location (the latter applying to the offset that is "ahead"), though 
"winter" and "summer" are also not bad, except that "summer" cannot be 
unambiguously abbreviated to "S". "Advanced" and "Basis" have the 
pleasing property of having initials "A" and "B", "A" time being ahead 
of "B" time. Should more than two offsets ever annually apply to a 
zone, subsequent letters of the alphabet could be used to distinguish 
them.

I hope I haven't strayed into territory outside the purview of this 
list.