chris-smith / rosforlabview Goto Github PK

When multiple publishers are run in a loop, only the first few will receive topic requests. The others will time out while waiting for a topic request after successfully registering. This issue does not seem to occur if the publishers are not run in the loop. Additionally, it is possible to start each publisher separately before running them in a loop, at which point all publishers work as they are supposed to. If the loop is stopped and topic servers are closed, the hanging topics will briefly connect before being shutdown. Possibly due to reentrancy on publisher specific vis in _BlankServer.

Temporary workaround is to start individual publishers before running them in a loop.

Labview fails to to open a TCP connection to a *.local IP address. This should only cause issues if the ROS Master IP is set to a *.local address, as we account for it when being given connection information from the Master.

There was an error with remapping arguments on the write call used by publishers where, if a space was found in the message but there was no valid argument, an empty string would be published instead of the message itself.

If two services are located at the same port and both are called, whichever service is connected first will probably maintain the connection. This is due to the current implementation in _BlankServer. LookupService is default behavior for service clients, and the connection to service is made immediately after this call. Nothing is done with it until the service is told to write something, when the connection is finally closed. Opening the connection should instead be done when the service is told to write something.

This issue occurs when making single calls to services, and causes an empty response to be returned. The issue is most likely due to timing issues between the write and read calls used for a service.

A publisher will occasionally lose its connection in the middle of a running program. Restarting publisher should fix the problem.