interactions.frame

Module nu.mem.interactions.frame.

Frame: a private mem store for one scope.

A frame is the stack frame of a Nu program. Whatever a body computes once and reads again, counts, or accumulates lives in the frame's own dict, laid out by a Shape, and goes away when the body ends. Nothing about it is new machinery: the dict is bound under (dict, Shape) with the scoped ctx.fabrics.bind every lifecycle bracket uses, and the Shape's mem refs read and write it like any other mem root.

Frame

Runs body with a fresh dict as the mem store for shape.

Frame(shape, body)

Path nu.mem.Frame. Kind Bracket, sort bracket, cardinality transparent. Arity 2 (2 required).

On entry the frame binds a new empty dict under shape and writes initial into its slots. The body reads and writes the Shape's mem refs against that dict. On exit the prior binding comes back and the dict is dropped, on a clean exit, an error or a cancel alike.

Arguments

Yields

Whatever body yields, in the body's own cardinality.

Notes

• Every run gets its own dict. Parallel arms, concurrent requests and nested calls each entering the frame never see each other's values; arms started inside one frame share its dict.
• A frame over a Shape shadows an outer frame over the same Shape for its body. Frames over different Shapes coexist.
• State that outlives one call, such as a throttle's last run inside a loop, needs the frame around the loop, not inside it.
• Reading a Shape's mem ref with no frame or binding for it in scope raises LookupError naming the Shape.
• An initial key that is not a slot of shape raises TypeError at construction. An initial value that evaluates to EMPTY raises on entry, as any mem write does.
• Over a stream body the frame stays bound for the whole drain.

Example

class Tally(nu.Shape):
    count = nu.IntRef.slot()
    total = nu.IntRef.slot()
step = Tally.count.set(Tally.count + 1) >> Tally.total.set(Tally.total + nu.Attr("item"))
mean = nu.ForEachDo(nu.Iter([1, 2, 6]), step) >> nu.print(Tally.total / Tally.count)
_ = nu.run(nu.Frame(Tally, mean, count=0, total=0))

3.0