On 8/11/22 05:39, Peter Krefting wrote:
The zone1970.tab files provides me with a list of time zones, one per country
Not sure about the "one per country" part. Some countries have more than one Zone, and some Zones have more than one country.
With the zone1970.tab change in 2022b, it is not immediately obvious that for Norway, I should select "Berlin" (although I have added the commentary, so it does say "Scandinavia").
Column 1 of the zone1970.tab entry for Europe/Berlin lists NO (Norway) as one of its country code, so if software looks at column 1 it should easily deduce that Europe/Berlin covers at least part of Norway. This sort of issue is not new to 2022b. For example, if software ignores zone1970.tab's column 1 it's not immediately obvious that Europe/Prague covers at least part of Slovakia, and this has been true ever since zone1970.tab was introduced in release 2014f.
I will have to experiment with changing the code to parse the zone1970.tab country list, and break it up by that, showing only the country name and the comment in the selector.
Yes, something like that needs to be done regardless of whether 2022a or 2022b is being used. In looking into this I did notice, though, that tzselect will be confused by 2022b's changes for Iceland, in that 2022b's tzselect doesn't know that Iceland is in the Atlantic, which means if you first select Atlantic, the secondary menu won't mention Iceland. Again, though, this sort of issue is not new to 2022b. For example, in 2022a tzselect if you first choose the Atlantic, tzselect doesn't know that St Helena is in the Atlantic and so its secondary menu won't mention St Helena. Here's another example: if you first choose Asia, 2022a tzselect doesn't know that Turkey is in both Europe and Asia and so its secondary menu doesn't mention Turkey. To work around this longstanding problem I installed the attached proposed patch, which adds comments to zone1970.tab to help tzselect avoid the problems mentioned above, and similarly for Cyprus, Svalbard, Mayotte, etc. I view these comments as experimental; if they work well we can promote them to full-fledged data in some future release and if not then perhaps we can think of something better.