You're creating a new VRF_NODE and installing it under CONFIG_NODE, but not installing any commands to VRF_NODE - those are going into CONFIG_NODE instead. So this is basically using it as a place to hook a vrf config-write function into the CONFIG_NODE config-write function.
What's wrong with it?
This kind of suggests the command node API would be better having a list of config-write function pointers in the cmd_node? Rather than installing dummy cmd_nodes with no commands (but commands added into the parent)? (I'm little bit of command.c spring cleaning that made me notice this).
OK, then, once there will be a "clean" logic, the "clean" rule could be followed. Currently, looking at the code, it seems either options are OK. I maybe missing something...
Thank you, Vincent _______________________________________________ Quagga-dev mailing list Quagga-dev@lists.quagga.net https://lists.quagga.net/mailman/listinfo/quagga-dev