holoscan::OutputPortConfigurator

Beta

template <PublishedPayload T>
class OutputPortConfigurator

Configure cold authoring policy for one Graph-owned output port.

A configurator is a non-owning handle. It remains usable only while the originating Graph and exact port declaration remain live. Each successful call atomically updates cold authoring state; a rejected call leaves the port and its incident edges unchanged. Configurators do not make Graph authoring thread-safe, so callers must serialize them with all other reads and mutations of the same Graph.

#include <holoscan/graph.hpp>

Template parameters

PublishedPayload

Exact published payload type of the output declaration.

Constructors

OutputPortConfigurator

holoscan::OutputPortConfigurator<T>::OutputPortConfigurator(
    std::weak_ptr<detail::GraphState> state,
    GraphPortId port,
    std::shared_ptr<const void> handle_identity
) noexcept

Methods

max_emits_per_compute

OutputPortConfigurator & holoscan::OutputPortConfigurator<T>::max_emits_per_compute(
    std::uint32_t count = 1
)

Override the positive per-compute emission bound.

Returns: This configurator for fluent authoring.

Throws: RuntimeError if count is zero or the source Graph is unavailable.

Parameters

count

std::uint32_tDefaults to 1

Positive maximum accepted emissions per compute call.

late_joiner_history_depth

OutputPortConfigurator & holoscan::OutputPortConfigurator<T>::late_joiner_history_depth(
    std::uint32_t depth
)

Override the finite late-joiner history depth.

Returns: This configurator for fluent authoring.

Throws: RuntimeError if depth is zero or the source Graph is unavailable.

Parameters

depth

std::uint32_t

Positive retained-sample depth.

produces_tensor

OutputPortConfigurator & holoscan::OutputPortConfigurator<T>::produces_tensor(
    const TensorOutputSpec &spec
)

Override this output with one Tensor spec.

kPassThrough does not remove Runtime-pool storage declared during Operator setup; Graph overrides may tighten setup resource declarations but do not retract them.

Returns: This configurator for fluent authoring.

Parameters

spec

const TensorOutputSpec &

Representation, finite bounds, and transfer promises.

payload_options

OutputPortConfigurator & holoscan::OutputPortConfigurator<T>::payload_options(
    PayloadOptions options
)

Override schema identity and payload options for this output.

Returns: This configurator for fluent authoring.

Parameters

options

PayloadOptions

Options copied before return.

diagnostic_alias

OutputPortConfigurator & holoscan::OutputPortConfigurator<T>::diagnostic_alias(
    std::string_view alias
)

Assign an owned diagnostic alias without changing port identity.

Returns: This configurator for fluent authoring.

Throws: RuntimeError if alias is invalid or the source Graph is unavailable.

Parameters

alias

std::string_view

Nonempty bounded UTF-8 alias copied before return.

state

std::shared_ptr<detail::GraphState> holoscan::OutputPortConfigurator<T>::state() const

apply

static void holoscan::OutputPortConfigurator<T>::apply(
    expected<void, Error> result
)

Types

Typedefs

Name	Definition	Description
`graph_authoring_handle_tag`	`void`	Opt this authoring handle out of replayable Graph captures.

Member variables

Name	Type	Description
`state_`	`std::weak_ptr< detail::GraphState >`
`port_`	`GraphPortId`
`handle_identity_`	`std::shared_ptr< const void >`