.Modelica.Utilities.System.getTime() returns bad year

[adrpo]

Running this model:

model GetTime
  Integer ms;
  Integer sec;
  Integer min;
  Integer hour;
  Integer mon;
  Integer year;
algorithm
  (ms, sec, min, hour, mon, year) := .Modelica.Utilities.System.getTime();
  .Modelica.Utilities.Streams.print("ms:" + String(ms) + "\n");
  .Modelica.Utilities.Streams.print("sec:" + String(sec) + "\n");
  .Modelica.Utilities.Streams.print("min:" + String(min) + "\n");
  .Modelica.Utilities.Streams.print("hour:" + String(hour) + "\n");
  .Modelica.Utilities.Streams.print("mon:" + String(mon) + "\n");
  .Modelica.Utilities.Streams.print("year:" + String(year) + "\n");
end GetTime;

with OpenModelica on Windows will give you year=10 and on Linux will give you year=9.
With Dymola 2019 on Windows ill give you year=10.

The code for the external C is here:
https://github.com/modelica/ModelicaStandardLibrary/blob/master/Modelica/Resources/C-Sources/ModelicaInternal.c#L1264
I haven't looked in detail into it, maybe somebody can spot the problem.

[adrpo]

Reference: https://stackoverflow.com/questions/58320730/modelica-print-current-time/

[HansOlsson]

The call should be:
(ms, sec, min, hour, mday, mon, year) := .Modelica.Utilities.System.getTime();

So that you get day of the month as well. I don't know why Linux results are from last month :-)

[adrpo]

Yeah, I read the doc wrong. But still the issue with the month on Linux remains.
Correct model should be:

model GetTime
  Integer ms;
  Integer sec;
  Integer min;
  Integer hour;
  Integer mday;
  Integer mon;
  Integer year;
algorithm
  (ms, sec, min, hour, mday, mon, year) := .Modelica.Utilities.System.getTime();
  .Modelica.Utilities.Streams.print("ms:" + String(ms) + "\n");
  .Modelica.Utilities.Streams.print("sec:" + String(sec) + "\n");
  .Modelica.Utilities.Streams.print("min:" + String(min) + "\n");
  .Modelica.Utilities.Streams.print("hour:" + String(hour) + "\n");
  .Modelica.Utilities.Streams.print("mday:" + String(mday) + "\n");
  .Modelica.Utilities.Streams.print("mon:" + String(mon) + "\n");
  .Modelica.Utilities.Streams.print("year:" + String(year) + "\n");
end GetTime;

[beutlich]

Hm, only for Win32 we increment the month

ModelicaStandardLibrary/Modelica/Resources/C-Sources/ModelicaInternal.c

Line 1306 in caaf8b1

tlocal->tm_mon = tlocal->tm_mon + 1; /* Correct for month starting at 1 */

[HansOlsson]

Hm, only for Win32 we increment the month

  ModelicaStandardLibrary/Modelica/Resources/C-Sources/ModelicaInternal.c


     Line 1306
  in
  caaf8b1





    
      
       tlocal->tm_mon  = tlocal->tm_mon + 1;     /* Correct for month starting at 1 */

And the same for the year as well, which will be even more obvious.

[beutlich]

Yes, it was wrong from the beginning when I copied it from Modelica_Noise for MSL v3.2.2

https://github.com/DLR-SR/Noise/blob/891468c5c25c39f49c785bbdb853a386a4d0e889/Modelica_Noise%201.0%20Beta.1/Resources/Include/ModelicaRandom.c#L208-L243

[sjoelund]

I guess the problem with easily assigning to the bad value is due to not returning enumeration (or wrapping it into a record).

[HansOlsson]

I guess the problem with easily assigning to the bad value is due to not returning enumeration (or wrapping it into a record).

Wrapping it into a record seems the most logical.

Alternatively I think that the risk would have been less if we had used the same order as IS 8601 and reversed the output order to return year, month, ... (but changing it now would be too confusing).

[adrpo]

Could we change it to a record for 4.x?

[sjoelund]

I think there might be a possibility to rename getTime() to Internal.getTime() or something with the same interface. And then create a new getTime() that works by calling that function and creating a record on top of it.

[beutlich]

I think there might be a possibility to rename getTime() to Internal.getTime() or something with the same interface. And then create a new getTime() that works by calling that function and creating a record on top of it.

This is addressed by #3247.