mpathsenser 1.1.3

This is a release with breaking changes due to removal of deprecated arguments. Please review
carefully before updating.

This release also supports changes from the new release of m-Path Sense (01/02/2023). Most notably,
the accelerometer and gyroscope are no longer samples of a continuous stream, but rather summaries
of these streams. Old versions are still supported by all functions.

Major changes

• Thanks to a new version of m-Path Sense, accelerometer and gyroscope have gained extra columns:
  • x_mean: The average acceleration or gyroscopic value along the x axis within a sample;
  • y_mean: The average acceleration or gyroscopic value along the y axis within a sample;
  • z_mean: The average acceleration or gyroscopic value along the z axis within a sample;
  • x_mean_sq: The square root of the x values within the sample;
  • y_mean_sq: The square root of the y values within the sample;
  • z_mean_sq: The square root of the z values within the sample;
    From these values, one could calculate the L1 norm and L2 norm like before.
• Added a new value timezone to all sensor data. Confusingly, this is not the timezone of the
  data itself (as this is always in UTC), but rather the timezone the participant was in at the time
  of the measurement.

Deprecations

• Removed deprecated parallel argument in fix_jsons(), test_jsons(), unzip_data(), and
  import().
• Removed deprecated overwrite_db and dbname arguments from import().
• Removed deprecated path and db_name arguments from copy_db().

Minor changes

• Provided support for dplyr 1.1.0.
• link() no longer adds an extra row before (if add_before = TRUE) or after (if
  add_after = TRUE) if the first or last measurement equals the start or end time respectively.
• Changed link_db() lifecycle status to deprecated as link_db() depends on link(). Eventually,
  link() might see changes in its functionality that will cause link_db() to break, so it is
  better to deprecate it already to motivate users to stop using this function.

Bugfixes

• Fix cross-reference to undeclared package ‘future’ in documentation.

• Fixed bug #8 where bin_data() incorrectly handled days occurring after DST change.

• step_count()

These functions delivered incorrect output and only allowed summaries by a fixed time frame, e.g.
by hour or day. These functions will be reimplemented (some with a different name) in mpathsenser
2.0.0.

Minor changes

• When add_before or add_after is TRUE in link(), no extra row is added if there already is
  a row with a timestamp exactly equal to the start of the interval (for add_before = TRUE) or to
  the end of the interval ⁠(add_after = TRUE).
• moving_average() now allows a lazy tibble to allow further computations in-database after
  having called moving_average().
• identify_gaps() is now slightly more efficient.
• get_data() is now case insensitive. In a future update, all sensor names throughout all
  functions will be made case insensitive.
• When using add_before = TRUE, link() no longer adds an extra measurement if the first
  measurement in the interval equals the start time of the interval exactly.
• get_data() now allows multiple participant_ids to be used.
• external_time has been added as an argument to link_db(), to be able to specify the time
  column in external_data in accordance with the change in link() above.

Bugfixes

• link() now correctly handles natural joins (when by = NULL) and cross joins (when
  by = character()).
• The column original_time was not added for any other nested data row except the first one,
  if add_before or add_after was true.
• link() no longer suffers from future's max object restriction (500MB by default).
• When x and y use different time zones in link() and add_before = TRUE, link() now
  correctly leaves all time zones equal to the input.
• link() incorrectly assigned the time zone of x to the nested data of y, if add_before or
  add_after was true. This is now changed to the time zone of y, to ensure consistency. Note that
  if the time zones of x and y are different, matching will be correct but the nested data may
  seem off as it will keep y's input time zone.