Gateway locations are not mandatory. However not delivering a location will cripple some features of the LoRaWAN network like metadata that comes with your received packet.
I am not sure I am understanding. Why do you need location for downlinks? For planning it could be partially useful, although a site survey is more suitable to have an idea of coverage.
Simple: The network server evaluates RSSI and SNR of all gateways receiving the node and selects the gateway with the best report for downlink of a packet.
I’d thought that the network used location in the scenario where a gateway had exceeded the legal transmission time.
E.g., node was heard by 1 gateway only A.
Gateway A has been transmitting and has used up all of its allotted transmission time.
Downlink is requested.
Gateway B is known to be geographically close.
I thought in this scenario the network attempted to downlink from Gateway B.
Are we saying that in this scenario the network just refuses to attempt to deliver the downlink?
If gateway B did not receive the uplink and forwarded it to TTN there is little chance a downlink sent by it will be received by the node. So not attempt to send a downlink is made.
Geographical proximity is not the same as “radio” proximity, which is better measured with RSSI and SNR. You might have a gateway in the basement that does not receive more than hundreds meters. Or, more realistically, you might have a node in a city that, due to lucky combination of buildings, trees, gateway altitude, etc, is received by one far gateway and not by a closer one.