Skip to content

wake-eval: activation latency is applied per-clip, not per-threshold #71

Description

@jamditis

Follow-up from the cloud review of #70 (P2). In wake_eval._activation_accepted a single stored Activation.latency_s is compared for every threshold, but the first-crossing time depends on the threshold. Inert today (driver builds Activation with latency_s=None); a design note for when the documented latency tolerance is wired to measure latency: latency must be the per-threshold first-crossing time, not a scalar. Ref: review thread on #70.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions