Curtis Vogt

> What about creating a simple binary serialization format just for this data then? The plan is to use the binary [tzfile](https://data.iana.org/time-zones/code/tzfile.5.txt) format. Maybe a slightly modified version to work...

> Seeing CI failures due to TZdata every once in a while. It would be really nice to get this done. I've been working on this issue during my free...

> I've been working on this issue during my free time as I've been unable to allocate working hours to this. Just to clarify to avoid any misinterpretation: this is...

A quick status update. As of TimeZones 1.8.0 we have support for reading/writing to the new TZJFile serialization format (this is based off of the tzfile format but with some...

I'm planning on releasing TimeZones 1.9.0 early next week, ideally Monday.

> I guess if TZJFile format works out in the future we could switch to pre-building the files and generating artifacts? Or does that not make sense? That is definitely...

Did some work over the weekend to get the 1.9.0 release ready. I've just [registered the release](https://github.com/JuliaRegistries/General/pull/64012) and it should be available shortly.

As `VariableTimeZone`s use `FixedTimeZone`s as part of the `Transition`s removing the `name` field would result in no longer being able to access the abbreviations names commonly used in North America.

I agree we can do better but I'd rather not lose information that is part of the tzdata. There are a few alternate ways we could be storing this data....

Unfortunately there are a variety of answers that could be generated when converting local `DateTime`s into `ZonedDateTime`. You can use the `occurrence` argument of the `ZonedDateTime` constructor to work around...