Pontifications

Following up from yesterday’s R and Lubridate: Calculating Time Differences in R using intervals aka a better way to compute daynumber starting from arbitrary day of the week

add_release_day_number <- function(
  df_release,
  yyyy,
  mm,
  dd
) 
{  
  START_DATE <- 
  make_datetime(
    yyyy, mm, dd, 0,0,0, 
    tz = "America/Vancouver"
    )
  
  return (
    df_release %>% 
      mutate(
        release_week_day_number = 
        floor(
            interval(START_DATE, created) / days(1)
            ) + 1
        )
  )
}

The above ^^^^ is the most accurate and the best way to compute daynumber starting at an arbitrary day of the week e.g. Tuesday for Firefox releases! You can also use lubridate durations, but durations aren’t accurate when you transition from month to month or year to year