So for a given foo.bar.com, the SRV check should be performed on _factorio._udp.foo.bar.com (since factorio exclusively uses UDP ports), and parse any found record for a port and host address to use during the connection.
This would make it far easier for server hosts to run multiple Factorio instances on one server and allow users to connect to them with distinct subdomains instead of having to enter port numbers.
For example, given a dedicated Linux server that has two headless Factorio games running, located at:
- _factorio._udp.foo.example.com. <TTL> IN SRV <priority> <weight> 34197 server.example.com.
- _factorio._udp.bar.example.com. <TTL> IN SRV <priority> <weight> 34198 server.example.com.
Minecraft's client for example has this capability: http://gaming.stackexchange.com/questions/106646 as of 1.3: http://minecraft.gamepedia.com/1.3.1#Gameplay_2 (search for "srv")