Thomas Eizinger

> @thomaseizinger > > > I think we need to embed the domain and proxy IPs in the initial upsert_flow message because this check is part of the authorisation and...

Summary from discussion with @AndrewDryga: We only ever want to authorise each resource once, even if it is just a wildcard resource, otherwise we'll see quite a lot of spam...

I'd like to avoid that because its only purpose would be to know explicitly, for how long we can buffer packets. It would add a delay of 2 RTTs though...

Thinking more about it, it might be better to buffer explicitly and wait for a response. That way, the experience is guaranteed to always be the same, regardless of the...

> I wonder if it'd be too much load on portal/gateways if we send all ips to all connected gateways as soon as we resolve them. I think it should...

I've thought about what the messaging between client and gateway could be here and I think it is possible but it isn't very clean and kind of brittle. The gateway...

> To replace that, we'd need to have a control protocol for the coms here in this issue, and then write glue code to tie the messages to the PhoenixChannel/LiveView...

> As we grow the team anyone seeing self-written transport protocol will be WTFing all the time :). At the moment, I am WTFing at our control protocol is complex...

@jamilbk @conectado Draft because I want to refactor a few things around mangled DNS packets and still need to fix the tests but would like to get a review on...

> I think it's worth the "perceived" UX improvement here to match competing products in our space if even only for `ping` IMO it is a bit more than just...