[tz] Why 2021b needed to be issued
eggert at cs.ucla.edu
Sat Sep 25 05:23:07 UTC 2021
Since 2013 I've been patching tzdb by replacing Zones with Links when
Links suffice according to the guidelines. These patches have not
affected timestamps after 1970. I've been doing this gradually. The
changes have worked without significant incident, partly because end
users generally don't care about pre-1970 timestamps, and partly because
the few users who do care (astrologers, mostly) know or should know that
the pre-1970 Zone data entries are often wrong and even when right are
typically grossly inadequate for realistic use outside the named
location - this is partly why any focus on pre-1970 is out of scope for
Preparing all these patches has been a big enough job that I've been
doing it gradually since I have limited time. This process was discussed
on the list before it began, and although there was some objection, a
consensus to move towards these goals was reached at the time. Most of
this job was done before 2021a came out.
The main difference in this round of patches, is that I had aimed to
complete the job, in response to a user query raising equity concerns in
2021a. It is clear that the 2021a setup was not sustainable in the long
term, if equity is a goal.
Completing the job generated larger patches than usual, and the patches'
size triggered concern that they held too many changes all at once.
Also, the patches' effects on pre-1970 timestamps in (for example)
Norway and Sweden have triggered concerns, much more so than similar
changes made (for example) to Angola and Congo in tzdb 2014g.
My own view is that we've done this before, many times, it hasn't been a
problem in practice, and it's way past time to address the equity
concerns as they have a good chance to really bite this project and its
repackagers in the future.
It's a bad look for us that so much concern about Norway and Sweden has
appeared on this mailing list, even though hardly anybody seems to have
cared about Angola and Congo. It'll be an even worse look if we ignore
this issue weeks, months or even years after it's been made clear to us.
I wrote, tested and published several concrete suggestions addressing
the fairness problem. In the months since May when the fairness problem
was discovered and objections to my proposed fixes were raised, no other
concrete fix addressing both of these concerns has been implemented or
tested, and all other suggestions made so far would have required
significant work, if they would have worked at all, or would have caused
more data churn than what I proposed.
I realize that there are valid concerns about installing too many of
these patches at once, so 2021b affects only nine Zones instead of the
thirty-odd Zones in my May proposal. The idea is to do further patch
sets in not-too-distant-future releases, so that the job can finally be
I also recognize that there are other forms of equity and fairness that
should be considered, perhaps along with alternative technical solutions
to the problems these patches address.
With all this in mind, issuing 2021b now is a significant step toward
equity in tzdb. It will let us say that we are moving toward a fair
process, and will give us the opportunity and motivation to improve on
that process and to address and balance the various other concerns that
have recently appeared on the mailing list.
More information about the tz