[Info-vax] EU will abandon daylight savings time in 2021

[Stephen Hoffman]

On 2019-03-28 21:55:03 +0000, John Reagan said:

> On Thursday, March 28, 2019 at 5:27:53 PM UTC-4, Dave Froble wrote:
>>> 
>> I think you totally missed my point Steve.

Wouldn't be the first time for that, either.

>> Why not let it up to the users?

Download and install their own timezone files?  That'd be an 
OpenVMS-classic solution.  Toss it over the wall to the users as 
another thing folks need to deal with.  Downside is that some number of 
folks are then going to call and be looking for support to do that.  Or 
they'll forget, some switch won't happen, and they'll call.

Or let the users set their own timezones?  That's possible (and I've 
done it), but that wasn't all that popular among the customers.

Or are we discussing users and organizations deciding when to start 
their own days?

>> Give them a tool to set whatever they want.

Users can set their timezone with the existing UTC support in OpenVMS.  
 And can create or alter their timezone definitions.

>> You're never going to make everyone happy, so let them do it.
> 
> You mean like the exiting LIB$DT_FORMAT support and the corresponding 
> SYS$MANAGER:LIB$DT_STARTUP.COM startup file?  Use DIR/DATE to see how 
> you can control that today.
> 
> And C programs can define TZ to pick their timezone.

The IANA / Olson database and the DTSS and C bits was all incorporated 
into OpenVMS at V7.3, rather than trying to deal with all the various 
pieces separately, and rather than trying to have the users deal with 
the timezones themselves.

So.... could you elaborate a little on your suggestion, David?






On 2019-03-28 21:31:56 +0000, Dave Froble said:

> UTC for every location, yeah, you'd need that.  But then the calc 
> should be rather simple.  Or am I totally missing something?

The calculation is simple.  The interval between two times depends on 
when you are, and also on where you are.   As well as some other 
details.  The associated data for these calculations is what is 
provided in the Olson database.

Sure, the simple cases work, and the calculations are easy.  This is 
also how we tend to get the time-related messes, when one of the 
developer's assumptions fails.

Some of those assumptions:

https://infiniteundo.com/post/25326999628/falsehoods-programmers-believe-about-time 

https://infiniteundo.com/post/25509354022/more-falsehoods-programmers-believe-about-time 


Here's one of my more "fun" examples: September 1752.  That month was 
19 days long in the United States, and 1752 was 11 days short.

https://www.timeanddate.com/calendar/custom.html?year=1752&month=9&country=1&typ=1&hol=0&holm=1&display=1&df=1 


For some calculations, leap seconds matter.  For others, not so much.

For OpenVMS, the dtss routines are probably the best resource, though 
there are other routines for calculating dates and durations.  The dtss 
API was integrated at V7.3 and was documented in the OpenVMS manuals in 
the early V8 range.



-- 
Pure Personal Opinion | HoffmanLabs LLC