Ignore Participant from LabVIEW Toolkit

[hackinzach]

I am wondering if it is possible for a DataReader to ignore DataWriters that share the same Participant ID?  An overly-simplified use case for this is to build an application that forwards traffic to/from another domain, similar to a Routing Service.  Yes, I'm familier with RTI's Routing Service and that has satisfied previous needs, but I need to be able to customize some of the processing within LabVIEW if possible.

Here is a simplified recreation of my problem: an application is subscribing to data on a topic on Domain 1, then broadcasting it out on Domain 2.  At the same time, it is also subscribing to the same topic on Domain 2 and broadcasting it out on Domain 1.  The problem occurs when the data forwarded to Domain 2 is being straight-away being read in as new data and sent straight back to Domain 1, creating a loopback that generates huge amounts of data.  See the attached "routingSimpleExample.png" to see a simplified Block Diagram that exemplifies this behavior (minus the loopback from Domain 2 to Domain1).

When this application is reading data from the topic on Domain 1, I want it to ignore any data published by its own Participant ID so I am truly just forwarding traffic from other participants through domains.  I found a forum post explaining how this could be done in C, C++, or Java using the ignore_participant function: https://community.rti.com/kb/how-do-i-get-datareader-ignore-datawriter-belongs-same-domainparticipan... but this isn't exposed in LabVIEW to my knowledge.  

I also tried using a combination of limiting the <transport_builtin> to UDPv4 (effectively turning off shared mem transport), while at the same time disabling the loopback interface by following these instructions in a QOS profile: https://community.rti.com/forum-topic/ignore-participant-dynamic-data  But this still didn't turn off the data from being read back.

Is there another way to accomplish this through QOS Profile updates?  I feel like I'm missing something simple.  Thanks in advance.

[angelmartinez]

Hi hackinzach,

(Correct me if I'm wrong) Currently we have this scenario:

• INPUT_DATA_1 (data) --> DR_1 (data) --> DW_2 (data) --> DR_2 (data) --> DW_1 (data) --> OUTPUT_DATA_1 (data)

Where DR/DW, means DataReader and DataWriter, and the number is the domain ID. The name in brackets is the topic name.

Then, the output_data becomes input_data again.

As you have mentioned, the LabVIEW API doesn't provide the `ignore_participant` behavior. However, there are a couple of ways that we can achieve this functionality. The implementation of these options may depend on your application configuration:

Option 1: Modify the topic name used

If the problem is just one machine that perform this task (writing to domain2 and reading back the data). We could modify the name of the data that comes from Domain 2, therefore, the Domain 1 Data Reader won't read this data. The rest of DataReaders should modify the topic name accordingly:

• INPUT_DATA_1 (data) --> DR_1 (data) --> DW_2 (data) --> DR_2 (data) --> DW_1 (DataFromDomain2) --> OUTPUT_DATA_1 (DataFromDomain2)

Then, the output data will be written in DataFromDomain2 topic and that prevents to use it as `INPUT_DATA_1 (data)`. The rest of applications should have a DataReader for that topic.

However, this doesn't work if we have several applications doing the same thing, because we will end up in the next situation:

• INPUT_DATA_1 (DataFromDomain2) --> DR_1 (DataFromDomain2) --> DW_2 (data) --> DR_2 (data) --> DW_1 (DataFromDomain2) --> OUTPUT_DATA_1 (DataFromDomain2)

And we create the loop again

Option 2: usage of content filters

We could add an additional field to the data that you are sending that will be 'source identifier'. Then we could filter out data that comes from the same 'source identifier'. The `source identifier` will be unique per application and will allow us to filter the data whose its `source identifier` is different to the `source identifier` of the current application.

For example, if we have one application whose `source_id = 1` (we could do this manually):

• INPUT_DATA_1 (data) [source_id = X] --> DR_1 (data) [source_id != 1] --> DW_2 (data) [source_id = X] --> DR_2 (data) [no content filter] --> DW_1 (data) [source_id = 1] --> OUTPUT_DATA_1 (data) [source_id = 1]

Definitions:

• [source_id = X]: source_id is the remote application source_id, in our case it will be always different than 1 because it is assign to the current application
• [source_id != 1]: content filter the data to read only samples show source_id is different than 1
• [source_id = X]: we keep the original source_id to send the data to the domain 2. We could also modify this value to be `source_id = 1`, this may depend on your application (see next point)
• [no content filter]: in domain 2, there is no restriction to read data from outside or from the same application, so there is not need to filter data. If this changes, we need to create a specific content filter and modify the source_id from DW_2 accordingly
• [source_id = 1]: assign the source_id = 1 to the produced samples in the Domain 2, that avoids to loop back the data in the DR_1

The second option will work for for all cases, but you need to modify the datatype to contain the source_id information and add content filters accordingly. This require that all source_id are unique per application. Note: you could use a different type for the source_id, such as the machine name (if each application runs in a different machine), or any other identifier that is unique for the application to avoid setting these values manually

I hope this helps, let me know if you have other questions