Join reader thread + serial conn validation check[CPP-593]

[notoriaga]

• Adds a join to the message reader thread.
• Adds a short one second validation of the serial connection to parse for a preamble message before starting up the process messages thread.

Here is a video showing what is observed connecting to a port with the incorrect baudrate. Note, there is a seven second debounce on the connection notifications which is why it does not show up each time. Looks like the first 10sec of the video are corrupted 😬 I cannot catch a break.

bad-baudy.mp4

[john-michaelburke]

@notoriaga I wasn't sure the best way to formulate my work in with yours so I pushed mine over yours. It does seem joining the reader thread was catching a corner case before disconnecting and attempting to connect to the same device with a different baudrate.

[notoriaga]

Something I thought of after we talked about this, what if the corruption still results in a preamble byte?

[john-michaelburke]

I guess my thinking was that if it shows up at all then it should hopefully show up again and once it hits the process messages loop the crc check should error out but I see how this is not the most deterministic behavior. However playing with this approach, it does seem to error out consistently for each of the available baudrates that are not expected. This might be something we would have to monitor on multiple devices.

Otherwise we would need to implement some additional logic in the parser. I guess we would not want to have the console dictate the behavior of the sbp parser. Either we could move some of the timeout logic out of the console and into the parser or potentially add a maximum amount of data parsed with no valid message before erroring out. 🤷

[notoriaga]

Makes sense, seems better than nothing 🤷

[john-michaelburke]

👀