A decorator that drives a whole cycle of the inner stage per outer run — like Repeat and Reduce, on the same outer input — folding every output the inner stage produces into an accumulator R via the binary operation stage op.

Unlike Reduce, the accumulator is never empty: apply receives the seed R alongside the input, so every output the inner stage produces — including the first — is folded via op((accumulator, output)); there is no seedless first output to special-case. Because the accumulator always has a value to yield, Fold mixes in h8io.stages.base.Fruitful: apply is statically guaranteed to return h8io.stages.Yield.Some.

That guarantee shapes what happens when op itself produces no output for a given fold (e.g. it filters some outputs out): the accumulator is left unchanged rather than discarded. Reduce can thread an unfolded value onward as None, since its accumulator is optional to begin with; Fold cannot — a Fruitful stage has no None to fall back to — so keeping the previous accumulator is the only option that preserves the guarantee. op's status is still combined into the iteration's status in this case, since op was applied (not skipped) for that output. When the inner stage itself produces no output, op is skipped instead — as in Reduce — and only the inner stage's status carries the iteration, so each iteration still performs exactly one of apply/skip on both inner stages, as the lifecycle contract requires. An iteration that leaves the accumulator unchanged keeps the cycle spinning as long as the combined status is h8io.stages.Status.Success, like Repeat. The cycle is @tailrec and does not grow the stack.

The cycle stops when the status of an iteration is h8io.stages.Status.Complete: an error-free Complete is reported to the enclosing pipeline as Success, a Complete with errors is preserved. The accumulator as it stands at that point is yielded. The accumulator is local to one outer run: the next generation of Fold starts fresh from whatever seed the next outer input supplies.

The inner pipeline

alterand and op together form the inner pipeline of Fold, with op downstream of alterand. An iteration's status is therefore the combination of their statuses in pipeline order (alterand first), and both continuations are selected by that combined status — exactly as h8io.stages.Evolution.AndThen evolves both sides of a composed pipeline with the status of the whole run, never with the stages' individual statuses.

As in Loop and Repeat, the continuations are never selected by the status Fold reports to the enclosing pipeline — the two differ, since an error-free Complete is absorbed into a Success — and for the same reason the returned evolution is a h8io.stages.base.ConstEvolution. Its dispose releases both inner evolutions via h8io.stages.Evolution.dispose (op first, then alterand), which per the h8io.stages.Evolution contract stay the terminal cleanup handles of the generation just constructed. Symmetrically a skipped Fold skips both inner stages and selects their continuations on the neutral Status.Success.