Errof for tsfresh features set

[MislavSag]

I am using the new version of the package, after some time and have problems with tsfresh features set.

I have converted a data into the tibble format.

       symbol                date  close
       <char>              <POSc>  <num>
    1:    SPY 2004-01-02 10:00:00 112.11
    2:    SPY 2004-01-02 11:00:00 111.90
    3:    SPY 2004-01-02 12:00:00 111.77
    4:    SPY 2004-01-02 13:00:00 111.70
    5:    SPY 2004-01-02 14:00:00 110.93
   ---                                  
31305:    SPY 2021-10-20 12:00:00 452.50
31306:    SPY 2021-10-20 13:00:00 451.80
31307:    SPY 2021-10-20 14:00:00 452.23
31308:    SPY 2021-10-20 15:00:00 452.40
31309:    SPY 2021-10-20 16:00:00 452.00

When calling the function

        theft::calculate_features(
          tsibble::as_tsibble(x_, key = "symbol", index = "date"),
          feature_set = params
        )

where params i tsfresh, I get an error

Running computations for tsfresh...
Feature Extraction:   0%|          | 0/1 [00:00<?, ?it/s]
Error in py_call_impl(callable, call_args$unnamed, call_args$named) : 
  numpy._core._exceptions._UFuncBinaryResolutionError: ufunc 'add' cannot use operands with types dtype('<M8[ns]') and dtype('<M8[ns]')
Run `reticulate::py_last_error()` for details.

I tried to coinvert datetime to date, just to check if it will work

> tsibble::as_tsibble(x_, key = "symbol", index = "date")
# A tsibble: 4,482 x 3 [1D]
# Key:       symbol [1]
   symbol date       value
   <chr>  <date>     <dbl>
 1 SPY    2004-01-02  112.
 2 SPY    2004-01-05  112.
 3 SPY    2004-01-06  112.
 4 SPY    2004-01-07  112.
 5 SPY    2004-01-08  113.
 6 SPY    2004-01-09  113.
 7 SPY    2004-01-12  113.
 8 SPY    2004-01-13  113.
 9 SPY    2004-01-14  113.

But than I get an error

Running computations for tsfresh...
Feature Extraction:   0%|          | 0/1 [00:00<?, ?it/s]
Error in py_call_impl(callable, call_args$unnamed, call_args$named) : 
  TypeError: unsupported operand type(s) for +: 'datetime.date' and 'datetime.date'
Run `reticulate::py_last_error()` for details.

I have installed the latest versoins of all packages (R and py)