diff options
Diffstat (limited to 'host/lib/include')
| -rw-r--r-- | host/lib/include/uhdlib/experts/expert_factory.hpp | 14 |
1 files changed, 12 insertions, 2 deletions
diff --git a/host/lib/include/uhdlib/experts/expert_factory.hpp b/host/lib/include/uhdlib/experts/expert_factory.hpp index 4568abd61..6821775e9 100644 --- a/host/lib/include/uhdlib/experts/expert_factory.hpp +++ b/host/lib/include/uhdlib/experts/expert_factory.hpp @@ -65,6 +65,16 @@ public: /*! * Add a expert property to a property tree AND an expert graph * + * The underlying property can be used like any other property tree property, + * including setting a coercer through set_coercer(). However, this means + * that the coercion is happening outside of the expert framework. This is + * primarily useful for tiny coercions (e.g., we accept both upper and lower + * case values, but only want lower case downstream) for which we don't want + * to bother with a full expert, or for which we don't want to trigger + * resolution at all (if mode is set to AUTO_RESOLVE_OFF). + * For more full-fledged coercion, prefer add_dual_prop_node(). This will + * properly engage the expert graph. + * * \param container A shared pointer to the expert container to add the node to * \param subtree A shared pointer to subtree to add the property to * \param path The path of the property in the subtree @@ -87,11 +97,11 @@ public: const auto_resolve_mode_t mode = AUTO_RESOLVE_OFF) { property<data_t>& prop = - subtree->create<data_t>(path, property_tree::MANUAL_COERCE); + subtree->create<data_t>(path, property_tree::AUTO_COERCE); data_node_t<data_t>* node_ptr = new data_node_t<data_t>(name, init_val, &container->resolve_mutex()); prop.set(init_val); - prop.add_desired_subscriber( + prop.add_coerced_subscriber( std::bind(&data_node_t<data_t>::commit, node_ptr, std::placeholders::_1)); prop.set_publisher(std::bind(&data_node_t<data_t>::retrieve, node_ptr)); container->add_data_node(node_ptr, mode); |