expose the connection kind to CustomClientAuthentication

[ripienaar]

Signed-off-by: R.I.Pienaar rip@devco.net

/cc @nats-io/core

[kozlovic]

I would approve, but knowing we are planning on a 2.2.2, a patch release, are we "allowed" to change APIs (even if in this case it is an addition, so I personally think it should be ok)?

[ripienaar]

I believe adding is fine, and also there's no way that we've been following semver moving from 2.1.x to 2.2.x so I think that ship has sailed imo

[kozlovic]

following semver moving from 2.1.x to 2.2.x

Maybe not, but there was no backward compatibility changes, if I am not mistaken, meaning that you can run a 2.2.x against a 2.1.x and vice versa. Of course some features are not enabled if not at the same level, but not a break like we had between 1.4 and 2.0 with routes, etc (hence the bump from v1 to v2). Anyway, I will LGTM it.

[kozlovic]

LGTM

[ripienaar]

The protocol did not change that is correct and we made efforts to make things like headers degrade gracefully(ish). But in numerous other data structures and behaviours I think were not quite up to semver standards.

We're following GoVer - do whatever it takes to avoid a major release :P