Re: [PATCH iproute2-next v6] ip-link: add support for nolocalbypass in vxlan

[Stephen Hemminger <stephen@networkplumber.org>]

On Tue, 23 May 2023 20:37:04 +0200
Andrea Claudi <aclaudi@redhat.com> wrote:

> Stephen, I'll try to summarize the discussion we had in v5 here.
> 
> - We agree that it's a good idea to have JSON attributes printed both
>   when 'true' and 'false'. As Petr said, this makes the code less error
>   prone and makes it clear attribute is supported.
> - I have some concerns about printing options only when non-default
>   values are set. Non-JSON output is mostly consumed by humans, that
>   usually expects something to be visible if present/true/enabled. I
>   know I'm advocating for a change in the iproute output here, and we
>   usually don't do that, but I argue there's value in having a less
>   cluttered and confusing output.
> 
>   For example, let's take what you see with a default vxlan:
>   $ ip link add type vxlan id 12
>   $ ip -j link show vxlan0
>   [...] udpcsum noudp6zerocsumtx noudp6zerocsumrx [...]
> 
>   IMHO printing only "udpcsum" is enough to make the user aware that
>   the "udpcsum" feature is enabled and the rest is off.
> 
> I'm not against Vladimir's change, of course. But I would be very happy
> if we can agree on a direction for the output from now on, and try to
> enforce it, maybe deprecating the "old way" to print out stuff step by
> step, if we find it useful.
> 
> What do you think?
> Andrea

If you look at the other RFC patch set. It does change to always
print the state of all options.