I guess the most performant way is to precalculate on update then? SQL-speak, there should be “tables” that keep running total for each granularity, e.g. total_in_month:
| User | Month | Count | Total |
|---|---|---|---|
| John | 2019-01 | 3 | 3 |
| John | 2019-02 | 2 | 5 (3+2) |
| Mary | 2019-01 | 2 | 2 |
Will there be any (performance) advantage of putting this kind of data in Dgraph as opposed to SQL?
Regarding GroupBy for date: Thanks for clearing it up!
All the while I thought @index(day) means “to index down to day granularity”, which means, I can query/group by year and month as well such as @filter(gt(date, “1980”)).