Paul, I view this as a style issue/discussion. I personally think any block that accepts an exit should have the exit so that humans reading the config file don't miss the context change.
As it's a style issue, I thought Donald's point on consistency across the CLI was valid. As I wasn't interested in fixing this *everywhere*, I didn't push on this. I'd still be fine with seeing the patch applied as is, for the topic to be addressed by someone in a cross quagga consistent way, or to decide it's not worth the discussion at this time. Thanks for asking! Lou On 5/26/2016 6:53 AM, Paul Jakma wrote: > On Tue, 5 Jan 2016, Lou Berger wrote: > >> >> On 1/5/2016 9:10 AM, Donald Sharp wrote: >>> I'm not sure that exit is the right way to go here, or at least I >>> would like some sort of agreement from everyone on what is the right >>> thing to write and then to actually make the cli entirely consistent >>> along these lines. >>> >> WFM >> >>> Can you help me get a deeper understanding on what the goal/motivation >>> of this patch is? Why only route-maps are we worrying about sub-modes? >>> >> I'm sure it's simply because we hit this case in one of our ad-hoc >> tests. So we simply fixed this case. >> >> Lou > Was there a conclusion to this? You re-subbed the patch after this > thread. > > What was the failing test? > > The config file parser doesn't need 'exit', but does a walk back up the > vty node tree (minimal as that tree tends to be) for non-matching > commands. > > Is this dealing with a problem in the quagga code, or a problem in > someone's test-harness? > > regards, _______________________________________________ Quagga-dev mailing list Quagga-dev@lists.quagga.net https://lists.quagga.net/mailman/listinfo/quagga-dev
