[tz] Policy Suggestion: Minimum time period betwen releases

Robert Elz kre at munnari.OZ.AU
Mon Mar 27 20:16:22 UTC 2023


    Date:        Mon, 27 Mar 2023 12:46:13 -0400
    From:        Howard Hinnant via tz <tz at iana.org>
    Message-ID:  <CB3E878E-1975-4529-892C-E5F95DC31E74 at gmail.com>

  | I propose a minimum period of 30 days between successive IANA tz
  | database releases.

That's a terrible idea, as has been said by others - but if any more
reinforcement were needed, consider what would have happened had Lebanon's
govt announced the original decision 2 days earlier, but otherwise everything
happened just the same as has occurred.

In that situation, 2023a would have included the change that had been
proposed - the proposed new rule would have had no effect, as the previous
release was almost 4 months earlier.   Further, there were changes to several
other countries in 2023a.

Now if the proposed rule were in place, there would be no opportunity to
fix it, once the change in Lebanon was altered (no matter what alteration
was made) for a month - beyond (or at least very close to) when applying some
other change would make any real difference.   Further, unlike now, when
(purely by chance) there is the option to simply ignore 2023b, that option
wouldn't really exist in the supposed case, or the other changes in 2023a
would be missed as well.   The only option would be for distributors, or
end users, to ignore the rule and apply their own changes (and hope to do
it correctly, which some doubltess could, others perhaps not).

In legal circles there is a well known saying "Hard cases make bad law".
What that means is that when one reacts to something going wrong with the
system, by a hasty attempt to change the rules (laws) to prevent that
situation arising, the results are almost always not what was intended.

Don't do that.

Similarly a rule requiring a delay between when governments announce
changes, and when they're added to the database and distributed, which
seemingly better, in that it might seem to require governments to act
more rationally - but isn't really, firstly because governments tend to
react badly to outsiders (like us) attempting to force some action (or
inaction) from them, and because of that, the only effect such a rule can
really have is to cause chaos inside the country, which the governemt would
blame us (or the sownstream software distributors) for.

The only thing we can do is attempt to educate and influence the decision
makers, and their advisors, of the impacts of these kinds of actions, and
attempt to convince them to make considered decisions, with plenty of
advance notice.

There's little chance of that being wildly successful, or not any time
soon, but there really is nothing else that can be done which is better
than what is currently done to handle this kind of situation.

kre



More information about the tz mailing list