currently EDF is based on time stamps and assumes a (mostly) uniform distribution of the data points. this works reasonably well for the sensor-based data such as heart rate. however, it is a poor fit for the data where we have long and potentially uneven distributions of data points, such as exercise classification (ran for 50min, walked for 4hrs, slept for 8hrs). if our goal is to cover this kind of data with EDF as well, then we probably need to adapt the format to also allow time ranges.
currently EDF is based on time stamps and assumes a (mostly) uniform distribution of the data points. this works reasonably well for the sensor-based data such as heart rate. however, it is a poor fit for the data where we have long and potentially uneven distributions of data points, such as exercise classification (ran for 50min, walked for 4hrs, slept for 8hrs). if our goal is to cover this kind of data with EDF as well, then we probably need to adapt the format to also allow time ranges.