[WIP] Improve handling of dynamic discovery - rmw_fastrtps changes

[arjo129]

Takes off from where #653 leaves off. Currently added support for IPv4 addresses as well as hostname based addressing.

Current status according to goal matrix (Ideal goal matrix can be found here, tools to compute the matrix can be found here):

Other Host
                           static defined       static not defined
                           1      2      3      1      2      3      default      
static defined     1       ✓      ✓      ✓      ✗      ✗      ✓      ✗      
                   2       ✓      ✓      ✓      ✗      ✗      ✓      ✓      
                   3       ✓      ✓      ✓      ✗      ✗      ✓      ✓      
static not defined 1       ✗      ✗      ✗      ✗      ✗      ✗      ✗      
                   2       ✗      ✗      ✗      ✗      ✗      ✗      ✗      
                   3       ✓      ✗      ✓      ✗      ✗      ✗      ✓      
                   default ✓      ✓      ✗      ✗      ✗      ✓      ✓      

[EduPonz]

Hi @arjo129,

A while back we had some conversations with @gbiggs and @clalancette about this. As a guideline, I prepared a PoC using a Fast DDS example that implements all this functionality. You can check it out in this branch. Important notes there are:

1. All the expected functionality works on that PoC.
2. It depends on a PoC implementation of DomainParticipant::ignore_participant that is not in master (although we plan to have it for Fast DDS v2.10 which in the intended Fast DDS version for ROS 2 Iron).
3. Fast DDS already resolves DNS names, so you don't need to worry about those.
4. At the time, I had to make some assumptions on how thing would be configured from a configuration formatting point of view, but I think you can adapt those easily.

Hope that helps!

[arjo129]

Hi @EduPonz, thanks for pointing me in this direction. I might be using the wrong branch for the tests. I'll take a look at the work done and see what needs to be done.

[arjo129]

Current status (some of the check marks are quite flaky, for instance a participant could be discovered but then message delivery does not take place):

☑️ means correct behaviour, ❌ means incorrect.

samehost
                               static defined                   static not defined
                               LOCALHOST  OFF        SUBNET     LOCALHOST  OFF        SUBNET     default
static defined      LOCALHOST  ✓          ✓          ✓          ✓          ✓          ✓          ✗
                    OFF        ✓          ✓          ✓          ✗          ✓          ✗          ✗
                    SUBNET     ✓          ✓          ✓          ✓          ✗          ✓          ✗
static not defined  LOCALHOST  ✓          ✓          ✓          ✓          ✓          ✓          ✗
                    OFF        ✓          ✓          ✓          ✓          ✓          ✓          ✗
                    SUBNET     ✓          ✓          ✓          ✓          ✓          ✓          ✗
                    default    ✓          ✓          ✓          ✓          ✓          ✓          ✗

Other Host
                               static defined                   static not defined
                               LOCALHOST  OFF        SUBNET     LOCALHOST  OFF        SUBNET     default
static defined      LOCALHOST  ✓          ✓          ✓          ✓          ✓          ✓          ✓
                    OFF        ✓          ✓          ✓          ✓          ✓          ✗          ✓
                    SUBNET     ✓          ✗          ✗          ✓          ✓          ✓          ✓
static not defined  LOCALHOST  ✓          ✓          ✓          ✓          ✓          ✓          ✓
                    OFF        ✓          ✓          ✓          ✓          ✓          ✓          ✓
                    SUBNET     ✓          ✓          ✓          ✓          ✓          ✓          ✗
                    default    ✓          ✓          ✓          ✓          ✓          ✗          ✗

[ros-discourse]

This pull request has been mentioned on ROS Discourse. There might be relevant details there:

https://discourse.ros.org/t/ros-2-tsc-meeting-minutes-2023-02-16/29927/1

[wjwwood]

After discussing it we decided to merge all of these into the original pr's and go from there. However, it's not clear to me if this replaces @gbiggs pr or if it replaces it. Currently this targets rolling and doesn't have commits from the other pr, so I'm assuming the latter, but then it needs rebasing.

[wjwwood]

Merging with #653