what does the nodeName of PublisherOptions/SubscriberOptions use for?

[qclzdh]

The example iceoption has one comment '// grouping of publishers and subscribers within a process" about nodeName, but that is not very clear to me, is there anyone can elaborate this?

I have several questions:
a. If a_publisher and a_subscriber is in same process, is it mandatory to use different nodeName?

iox::popo::Publisher<RadarObject> a_publisher({"Radar", "FrontLeft", "Object"});
iox::popo::Subscriber<RadarObject> a_subscriber({"Radar", "FrontLeft", "Object"});

b. If b_publisher and b1_publisher is in same process, do they need different nodeName, and what is the difference?

iox::popo::Publisher<RadarObject> b_publisher({"Radar", "FrontLeft", "Object"});
iox::popo::Publisher<RadarObject> b1_publisher({"Radar_b1", "FrontLeft_b1", "Object_b1"});

c. If c_subscriber and c1_subscriber is in same process, do they need different nodeName, and what is the difference?

iox::popo::Subscriber<RadarObject> c_subscriber({"Radar", "FrontLeft", "Object"});
iox::popo::Subscriber<RadarObject> c1_subscriber({"Radar_c1", "FrontLeft_c1", "Object_c1"});

[mossmaurice]

@qclzdh Good question! As written in the in the comment, a node is used to sort the Publisher's and Subscriber's inside a process. The concept of a node is derived from ROS 2. Other frameworks like AUTOSAR use the term "Runnable". Usually those refer to sub-part of an algorithm. The name is then used e.g. in the rmw_iceoryx implementation when creating node instances.

a. If a_publisher and a_subscriber is in same process, is it mandatory to use different nodeName?

This is up to you and depends on how you want to structure your software.

b. If b_publisher and b1_publisher is in same process, do they need different nodeName, and what is the difference?

The publishers and subscribers won't behave differently. You will see a different value in the column "Node" in iox-introspection-client.

I think it's a good idea to mention the node more prominently in one of the examples. What do you think @elBoberido @MatthiasKillat ?

[qclzdh]

Thanks~

[elBoberido]

@mossmaurice I'm also not quite sure about the purpose of the node. I know that it changed some time ago but can't tell what it is supposed to do now. Maybe @budrus or @FerdinandSpitzschnueffler knows more

[budrus]

@qclzdh @mossmaurice @elBoberido

The node names are only an optional thing for grouping and do not affect the functionality. We introduced them for ROS 2, there each publisher and subscriber belongs to a node and for the debug tools you need this information. We thought it might also be helpful for other frameworks that come with a component model like the nodes in ROS. If you have several of these components inside a process they may have publishers and subscribers for the same topics. Then it is good to have another dimension to differentiate them in our introspection client.