topic How does the tailf:action extension differ from the yang rpc statement? in NSO Developer Hub Discussions

How does the tailf:action extension differ from the yang rpc statement?

martmitc — Fri, 01 Mar 2019 12:05:00 GMT

As per the title, trying to understand is these are equivalent

rpc doSomething {

input {

leaf action {

type string;

}

output {

leaf dummy {

type string;

}

and

tailf:action doSomething {

input {

leaf action {

type string;

}

output {

leaf dummy {

type string;

}

Re: How does the tailf:action extension differ from the yang rpc statement?

mmalysz — Mon, 22 Jan 2018 21:27:00 GMT

The difference is that rpc is defined at the top level in the data model. That is how YANG RFC defines it. Now tailf:action is an extension that allows you to put rpc into any level in the data model.

Re: How does the tailf:action extension differ from the yang rpc statement?

martmitc — Tue, 23 Jan 2018 07:44:34 GMT

Thanks for the clarification, so any northbound system interacting with a device using the tailf:action would also need to support this behaviour, as it is not just something internal to the device implementation.

Re: How does the tailf:action extension differ from the yang rpc statement?

mmalysz — Tue, 23 Jan 2018 12:59:03 GMT

Well for the NB system it is matter of just providing a right context. For example in case of RESTCONF the only difference will be the URL. In both cases it will a POST method and the URL will differ - it will be simply a path to the action point

Re: How does the tailf:action extension differ from the yang rpc statement?

martmitc — Tue, 23 Jan 2018 13:25:43 GMT

I was thinking more in the case of the northbound using NETCONF, as agree it doesn't really make much difference for RESTCONF.

Re: How does the tailf:action extension differ from the yang rpc statement?

Jan Lindblad — Thu, 25 Jan 2018 09:01:00 GMT

Yes, the agent would need to know what tailf:action is in order to use it. Note that in YANG 1.1 a new keyword "action" has been defined (without the tailf: prefix) as an alternative to rpc, and works the same as tailf:action. Now that all major toolchains support YANG 1.1, there is no reason to use YANG 1.0 for any new work.