Juergen Schoenwaelder <j.schoenwael...@jacobs-university.de> wrote: > On Fri, Dec 08, 2017 at 07:08:32AM -0500, Lou Berger wrote: > > On 12/08/2017 06:15 AM, Juergen Schoenwaelder wrote: > > > > > >> In talking to some others on this topic, they suggested using a library > > >> per > > >> datastore. I haven't look into this enough to know if that is a good or > > >> bad > > >> idea, but it seems functionally equivalent to your first option but > > >> realized > > >> in a different way. Just something to add to the mix. > > > If people want to put additional options on the table, they should > > > work out the details and write down a tree diagram and send the result > > > to the list. > > > > I guess we have a different philosophical approach here. I'd prefer to > > hear about fresh ideas, even if half baked or asked as a "stupid > > question". Sometimes, if not dismissed out of hand, these lead to > > something that is a better result than others that have been immersed in > > the issue have thought of. > > > > The devil is usally in the details and 'using a library per datastore' > is for my taste a bit too little to understand what is actually being > proposed. I am not dismissing proposals, but I like to be able to > understand what is being proposed. And for that I need a proposal that > is a bit more than 'using a library per datastore'.
I agree. In this case, I can't understand how it would work at all. The library is config false, so a client can never get it via get-config from <running>. Maybe the proposal is a new operation <get-yang-library> with the datastore as a parameter? Or maybe the idea is to somehow return meta data together with <get-config>? I can guess, and dismiss the proposal based on my guesses ;-) Or someone can write up a concrete proposal. /martin _______________________________________________ netmod mailing list netmod@ietf.org https://www.ietf.org/mailman/listinfo/netmod
