Clarification to the activity_occurrence column

[tommiranta]

In the specification you state that:

The column activity_occurrence represents the number of times a given entity has done that activity, at the time that activity occurred (starting with 1). activity_repeated_at is the timestamp of the next time that same activity occurred for that entity. Both should be NULL if another activity has not yet occurred.

I think that there is a discrepancy between above statement and proposed usage of the activity_occurrence column in the examples section:

They can be used to easily get the first time each customer did an activity (activity_occurrence = 1) and the last time (activity_repeated_at = null). Since these two expressions each return only one row per customer, they're also a very efficient way to get every unique customer that has done an activity.

You'd have to query activity_occurrence = 1 OR activity_occurrence is NULL in order to get the first activity for when there is one or several activities of the same kind for the same customer?

I would appreciate a clarification to this issue. Is it a bug in the spec or something that I do not understand?

[tommiranta]

I'll further clarify how I interpret the situation. When building an activity schema implementation:

1. Both fields are NULL when we run transformation SQL queries and insert the results into the activity stream
2. When we scan the activity stream and fill in the activity_occurrence and activity_repeated_at columns
   • Increment the value of activity_occurrence. Set to 1 if there is only one activity of this kind for the customer
   • Set the ts value of next activity of the same kind to activity_repeated_at leaving the last activity with a NULL value