@Paul and Matt

Thanks both for answering.

@Matt, when i try to update tzdata it says it is already up to date. My impression was that it was not, because of the Morokko 2018 switch. But that assumption seems false. I read you are still debating about Morokko.

@Paul, it's a long story and an important one for me because many people find it unforgiving if software in my field has errors in it's timezone atlas. Since i discovered the trick with zdump it's a lot better but not perfect. As said, America/Chicago is notorious, over the years i had many reports of wrong timezones there, a notable one is the birth of Dylan, Bob 24 May 1941 21:05:00 Duluth MN, United States, 92w06, 46n47, America/Chicago.
The timezone as listed on astro.com (they are regarded as the standard just as the people from astrolabe) is 6 hours (https://www.astro.com/astro-databank/Dylan,_Bob).

But the database gives 5 hours.
/usr/share/zoneinfo/America/Chicago  Sun Apr 27 07:59:59 1941 UT = Sun Apr 27 01:59:59 1941 CST isdst=0 gmtoff=-21600
/usr/share/zoneinfo/America/Chicago  Sun Apr 27 08:00:00 1941 UT = Sun Apr 27 03:00:00 1941 CDT isdst=1 gmtoff=-18000
/usr/share/zoneinfo/America/Chicago  Sun Sep 28 06:59:59 1941 UT = Sun Sep 28 01:59:59 1941 CDT isdst=1 gmtoff=-18000
/usr/share/zoneinfo/America/Chicago  Sun Sep 28 07:00:00 1941 UT = Sun Sep 28 01:00:00 1941 CST isdst=0 gmtoff=-21600

There are many more examples like this. I regret not having collected them else i could have listed them, one more recent example is this birth given to me by the man's wife.
23/07/1963 20:03:00 Conway AR, United States, 92w27, 35n05, America/Chicago
Same case, it should be 6 hours but it's listed as 5.

Lately i was mailing with a collegue and for what it's worth, he says he also felt America/Chicago is notorious, he told me he had been investigating time zones on his own in various states in that area and made many corrections. As this is just hear-say i can't comment on the validity of his claims, but my experience is along the same.

I regret not being able to present more data to backup my experience, and i sincerely hope this is not seen as criticism to your excellent work.

Thanks for your time.
Jean

On 12/11/2019 21:23, Paul Eggert wrote:
On 11/9/19 3:51 AM, Jean via tz wrote:
There are other problems, like the America/Chicago timezone which is notorious for giving false results for years before 1970

Although Matt addressed your other points, I'm curious about this one. In what sense is tzdb's America/Chicago "notorious" for being wrong? I used Google to search for '"America/Chicago" false results' and the only hits I came up with were downstream bugs (e.g., in PHP installation, or in mistaken use of Python primitives), not bugs in tzdb itself.