I’ve been following this thread a bit. 

One thing I’d note is that there could be a difference based on time zone if 
the date is being parsed as a timestamp in the local time zone.

I believe I saw Patrick’s time zone as Central European (UTC +1:00) earlier in 
the thread. I didn’t mention this earlier because I assumed Riccardo was in the 
same time zone (Italy is also Central European Time Zone IIRC). But, it is 
always possible that Riccardo is using an alternative TZ on his computers such 
as GMT or UTC? I do this with all of my servers. Only my personal laptop is set 
to the local time zone.

Anyway, if you parse a date into a Unix timestamp in the local time zone, 
assume the time portion to be “all zeros”, and are east of UTC +0:00, the “day” 
component when stored as offset from January 1, 1970, will be the previous day. 

For example, today is my lovely wife’s birthday (November 18). If your local 
time zone is ahead of UTC (e.g., UTC+08:00) and you try to parse her birth date 
in local time zone and assume 00:00:00 as local time, this is actually a time 
on the previous UTC day. The time 2025-11-18 00:00:00 in a UTC+08:00 time zone 
is 2025-11-17 16:00:00 in UTC.

For something like a birth date, I think the parsing and the formatting should 
always be done with respect to UTC +0:00 time zone to prevent these bogus “date 
shifts”. I’m nowhere near a computer (responding from my phone) to look at 
either GNUstep or the AddressManager source code, but I think it would be worth 
it to spend a little time exploring whether this is a possible issue. 

HTH

—Robert 

> On Nov 18, 2025, at 17:47, Riccardo Mottola <[email protected]> 
> wrote:
> Hi Patrick
> 
> Patrick Cardona wrote:
>> Using latest revision 4112 from savannah, I got again the importing
>> issue.
>> 
>>> Please try John Doe! (Attached)
>> 
>> See my screenshot below: BDAY of John Doe is not as expected:
> 
> on one side, I belive your screenshot. On the other side I am pretty
> sure I did not introduce regressions lately. You reported sometimes
> working and sometimes not.
> 
> I updated my other workstation running FreeBSD and latest gnustep and
> AddressManager and tried there. It works.
> Attached the screenshot of exactly the same VCF I posted to the mailing
> list.
> 
> So I think it can be:
> - a real issue on your side. In that case I can only think on your
> locale. Would you mint trying to set e.g LANG=C.UTF-8 as a test?
> - an issue on your build side, Maybe you have two versions installed
> (System vs. Local) or some other "dirtiness"
> 
> Or well, something I cannot think of, including a issue with base.
> 
> Riccardo

Reply via email to