Version 4.5.5 is online
version 4.5.5 (26/10/2020)
- Website: Fixed bug for DST switch. Station time now is correct (not sure it is for some pacific/Australian situations where minutes start to count).
- Monthly logfiles: Fixed bug for existence of AirLink logfiles.
Now those files are ignored while reading the monthly logfiles.
NOTE: @meteosangonera: testing of the dayfile issue still waiting for the new CMX release, we need some patience. And in addition, your cumulusutils.ini file probably contains a section [website] (even if you don't use that) with parameters
ShowUV=true and
ShowSolar=true. Accidental they cause a generation of the solar graphs and should be set to false.
NOTE: I was a bit confused by the times some of the sites (v 4.5.4) displayed, because they were correct which algorithmically was not possible. This means that the user must have modified either the local time or the timezone on the machine which runs cumulusutils (or modified the resulting code). Worst case is that I do not understand what is going on which is an option. Another option is that it is a difference between RPi and Windows machines (running CMX/CumulusUtils) in which case I am not sure it is the end of the issue. I prefer getting a message if something is wrong.
In summary the times must always be as follows:
- Station time is always the actual time of the station with or without DST, in my case UTC -1 (or -2 in DST). This time is always more or less equal to the last update time.
- Browser time always shows the time of viewer of the site. So, if somebody in the UK is now watching my site, a time of one hour less than my time should be observed (UK is now on UTC)
- UTC is displayed for reference and is what it is.
Times should be correct always without modifying your computer time.
Automatic and correct time setting on the machine which runs CMX/CumulusUtils is assumed.
RESULT:
The clock on the dashboard for
Station Time of the generated website did not react well on DST change back to wintertime in some instances. The 4.5.5 bugfix release was installed on both Windows and RPi's and this showed that the cause is in the difference in handling of timezone differences on Windows and RPI (native .NET and Mono).
This original line of code works in Windows (native .NET) only:
Code: Select all
(TZ.GetUtcOffset(DateTime.Now) - TZ.BaseUtcOffset).Hours
And this fixed line works on native .NET and Mono:
Code: Select all
TZ.GetUtcOffset(DateTime.Now).Hours - TZ.BaseUtcOffset.Hours
Now lets see how this behaves when DST is activated again in March.
This had effect, not only on the clock, but also on the rise/set times of sun and moon.
So: Users with CMX/cutils running on RPi really need the 4.5.5 version.
The documentation shows that the calls used are introduced in various versions of .NET (and thus Mono) so the runtime version installed may also affect the result.
Always use .NET/Mono to the latest possible update level.
. 
).