Skip to main content

Aggregates

Aggregate functions start with an underscore. They're called like methods after a table or column.

Counting​

shop.orders._count()
→ 42

Aggregates return scalars — you don't need a projection.

Sum, avg, min, max​

shop.orders._sum(total)
→ 1842.50

shop.orders._avg(total)
→ 43.87

shop.orders._min(total)
→ 9.99

shop.orders._max(total)
→ 299.00

Arithmetic on scalars​

Aggregate results compose with + - * / % **:

shop.orders._count() + 5
→ 47

shop.orders[status = "paid"]._sum(total) / shop.orders._count()
→ 43.87

Useful for dashboards where you need a quick derived number without client-side math.

With a filter​

Filters compose with aggregates exactly the way you'd expect:

shop.orders[status = "paid"]._sum(total)
→ 1842.50

Sorting​

_asc and _desc are sort aggregates that flow into a projection or slice:

shop.orders._desc(placed_at)[0:10]{id, total, status}

Reads: orders sorted by placed_at descending, take the first 10, return these fields.

Response:

[
  { "id": "o99", "total": 299.00, "status": "paid" },
  { "id": "o98", "total": 45.50,  "status": "pending" }
]

Distinct values​

shop.customers._unique(country)
→ ["IN", "US", "DE"]

Pattern match (LIKE)​

Although _like is technically an aggregate, it's almost always used inside a filter:

shop.customers[name._like("Ada%")]

Date parsing​

shop.orders._date(placed_at, "2006-01-02")

The format string follows Go's reference layout.

What's next​

The big one — context-aware queries, the feature that lets you delete your API layer.