Input Parameters

Every query so far has had its filter values written directly into the query. The :in clause lets you declare named inputs, so the same query can be run with different values, exactly like function parameters.

Every one of those earlier queries actually had an implicit :in $: the database itself, conventionally named $. When you write :in yourself, you list $ explicitly, followed by your own inputs, in the same order the query expects them. The site passes $ for you either way. In Clojure you would call a query with one extra input like this:

In the editors below, the inputs are written after the query, one per line, in the same order as :in (after $). You can edit them and run the query again.

Single scalar

A bare variable in :in binds a single value. This is the most common form. Here ?type is bound to "Fire":

You can declare as many inputs as you need. Pass one value for each, in order. If a value is missing, Datomic tells you how many inputs it expected. That count includes $, so a query with one input of its own expects 2.

Collection binding

Wrap the variable in [?var ...] to accept a collection of values. The query matches any entity whose attribute value appears in the collection, like SQL's IN (...):

Tuple binding

Wrap multiple variables in [?a ?b] to destructure a single input tuple. Useful for passing a pair of related values, like a range:

Relation binding

A relation binding accepts a collection of tuples, like a small in-memory table. Each tuple is destructured into the declared variables. Here every row pairs a Pokemon's name with the least HP we want to see for it:

Each row in the input relation acts like a separate filter applied in parallel, then the results are combined.