[tz] the tzdb information schema

Paul Eggert eggert at cs.ucla.edu
Mon Jul 4 14:28:59 UTC 2022


On 3/25/22 15:36, Michael H Deckers via tz wrote:
>
>    • The fact that two timezones agree since 1970 is not represented
>      independently from the data of each timezone.

Yes, and more generally the problem is that the zic input format does 
not allow one Zone to say something like "before 1966 I was like this 
other Zone", or "between 1922 and 1945 I was like this other Zone". This 
problem is not limited to data in the "backzone" file.

To fix this more general problem, we could change the zic input format 
and change the data accordingly. (This has already been proposed, but 
has not been implemented.) Or we could have some sort of prepass over 
the data.Or we could do both. I doubt whether it'd be worth the hassle 
of trying to fix it only for "backzone".


>
> The file backzone is an integral part of the tzdb data,
>      not just a container for additional data of lesser quality.

Here I'm afraid I'll have to disagree. "backzone" contains a 
considerable amount data of lesser quality. (Some of the lack of quality 
is that we don't record or even know how low the quality is.) I don't 
have time to maintain "backzone" well and I doubt whether anyone else 
does either.

This need to limit the maintenance burden (much of it political, and 
some behind the scenes) is not something always appreciated by users. 
That doesn't make it any less real.


>
>      Thus we should get rid of Zone data for Argentina/Rosario etc
>      (or else update them); keeping data that are known to be wrong
>      is not only useless, it is an invitation for consequential errors.

Feel free to propose changes to "backzone" along these lines. Please 
send them in "git format-patch" form. If there's no objection it 
wouldn't be much work to install them.


>
>    • The fact that two timezones agree since 1970 has nothing to do
>      with the fact that some timezones have changed their names, with the
>      old names being kept as Links to the new names. Currently, however,
>      Links representing one or both facts are kept in the same file 
> backward,
>
Yes, that's a problem and should be fixed. Much of it is a relic of last 
year's controversy, which ended up with only nine zones being moved to 
"backward" instead of the thirty-odd that I originally proposed, under 
the idea the original proposal was too big a change to install all at 
once. It's time to move the rest of the zones (or at least, move nine 
more), and once the move is done we can attack the problem of 
categorizing Link lines better than they're categorized now, with the 
goal of making it easier for people like Stephen Colebourne to maintain 
downstream versions that use a different approach.




More information about the tz mailing list