Improving fjagepy

[notthetup]

Improvements to fjagepy coming in mainly from issues/warnings found downstream in https://github.com/org-arl/unetsockets/tree/master/python/unetpy

• Updated the _msg_rx method in Gateway.py to ignore incoming messages with the exact content {"alive": true}, preventing unnecessary processing of alive messages.

• Refactored the __str__ and _repr_pretty_ methods in Message.py to provide clearer and more informative string representations of messages, including handling cases with empty content and improving pretty printing.

• Added an agentTypes field to the JSONMessage class to support lists of agent types in message payloads.

[mchitre]

Are we sure we should not respond to alive messages in fjågepy?

[mchitre]

Are we sure we should not respond to alive messages in fjågepy?

[notthetup]

Are we sure we should not respond to alive messages in fjågepy?

I guess we missed capturing this in the Gateway spec. Who's the first to send {alive: true}? The master container or the gateway? I am currently under the assumption that it's usually Gateway that sends it and the Master container replies?

All we currently mention is

Must send a {"alive": false} message to the master container before closing.

https://github.com/org-arl/fjage/blob/1a87755eb9688bdfc743631ba2efaec9f4735b20/gateways/Gateways.md#close--void---void

[mchitre]

Are we sure we should not respond to alive messages in fjågepy?

I guess we missed capturing this in the Gateway spec. Who's the first to send {alive: true}? The master container or the gateway? I am currently under the assumption that it's usually Gateway that sends it and the Master container replies?

As soon as a slave connected, master says:

{"alive": true}

If the slave does not respond to this within a timeout (15 seconds or so), master drops the connection.

[notthetup]

Most Gateways start by sending a {alive: true} at connection , which seems like enough in current implementations?

[mchitre]

Most Gateways start by sending a {alive: true} at connection , which seems like enough in current implementations?

Yes, I believe so. Because this validates the gateway as alive.