fluidb/PLAN.md

Operations on relations

Operations

• from(...rels), can use chain notation
• left_join(left, right, ...conds), similarly for right_join, outer_join)
• concat(...rels)
• intersect(...rels), similarly for union
• diff(left, right), similarly for sym_diff
• select(rel, binding: {..})
• where(rel, ..conds)
• take(rel, n)
• skip(rel, n)
• sort(rel, expr1, expr2: sort_dir)
• group(rel, binding: {*key1: expr1, val1: expr2}, *ordering) may order elements within groups
• walk(pattern, ...conds, ...bindings)
• walk_repeat(pattern, ...conds, ...bindings) every element contains additional _iter and _visited fields
• values(data, ?Table)
• nested_values(data, ?Table).extract(Table)
• update(rel, Table)
• delete(rel, Table)
• insert(rel, Table)
• upsert(rel, Table)

Helpers

• print(rel)
• print_schema(rel)
• print_plan(rel)
• print_optimized(rel)

Aggregation

• Aggregation functions should implement .step() and .result()

Differentiation

• function calls use parentheses, names start with lowercase letters or "_"
• aggregation calls are the same as function calls except that square brackets are used instead
• query calls are the same as function calls except that query names start with upper case or "#"

Others

• assoc magic better work
• language within query
• constraints and indices
• datetime and array types
• GUI and TUI
• query optimization
• regex as a value type

Walk(a:A-[e:E]->b:B,
     a => (a.id == 10),
     e => (e.id < 20, count[] <= 1, rand() => asc),
     b: {
        ...b,
        ...e,
        ...a
     })

Chain(a:A-[p:Path]->b:B,
      _next => _hops < 10)