1. If the destination has a waypoint proxy, we must send to it to the waypoint (using HBONE).
   When we do this, we will want to preserve the original destination Service IP, as the waypoint can do a better job picking a backend pod than we can.

    could use new flow control capacity before acknowledging it, causing strict HTTP/2 servers to
    fail the call.


## Version 4.1.1

_2019-09-05_

 *  Fix: Don't drop repeated headers when validating cached responses. In our Kotlin upgrade we
    introduced a regression where we iterated the number of unique header names rather than then

		if !HasPrefix(o.Marker, o.Prefix) {
			return entries, io.EOF
		}
	}

	// With max keys of zero we have reached eof, return right here.
	if o.Limit == 0 {
		return entries, io.EOF
	}

	// For delimiter and prefix as '/' we do not list anything at all
	// along // with the prefix. On a flat namespace with 'prefix'
	// as '/' we don't have any entries, since all the keys are
	// of form 'keyName/...'

- **Reduce the size of the Gradle distribution** - We no longer need to carry multiple standard libraries for different languages.

## Decision

We do not introduce new public API methods that include Groovy types in their signatures.
Existing Groovy methods will not be removed immediately.

## Status

ACCEPTED

## Consequences

		cfg:    cfg,
	}

	// By detecting iptables versions *here* once-for-all we are
	// committing to using the same binary/variant (legacy or nft)
	// within all pods as we do on the host.
	//
	// This should be fine, as the host binaries are all we have to work with here anyway,
	// as we are running within a privileged container - and we don't want to take the time to
	// redetect for each pod anyway.
	//
	// Extreme corner case:

	switch event.Event {
	case controllers.EventAdd:
		// pod was added to our cache
		// we get here in 2 cases:
		// 1. new pod was created on our node
		// 2. we were restarted and current existing pods are added to our cache

		// We have no good way to distinguish between these two cases from here. But we don't need to!
		// Existing pods will be handled by the dataplane using `GetAmbientPods`,

                   If we're building with JDK8, then clearly we're sticking to
                   JDK8 APIs. However, I assume (but did not confirm) that we'd
                   have the same issue with our -android builds, which need
                   animal-sniffer so that they can check that we're sticking to
                   JDK6-like APIs.

                   So instead, we use javac9, which doesn't lead to this error.

## Consequences

* X
* Y
* Z
```

## Status

ACCEPTED

## Consequences

* We start to use Architectural Decision Records
* We use the proposed template from this ADR
* We locate `.md` files in the folder `/architecture-standards`
* We highly encourage usage of ADR to communicate decisions

## Decision

We do not use Java serialization.
Instead, we use custom serialization where we explicitly describe how data objects should be serialized and deserialized.

For internal purposes, we use binary formats for their brevity.
We use the `Serializer` abstraction to separate the actual implementation of serialization from its uses.

When sharing data with external tools, we use JSON.

## Status

ACCEPTED

            <!-- Even after we stop using JSR305 annotations in our own code, we'll want this link so that NullPointerTester's docs can link to @CheckForNull and friends... at least once we start using this config for guava-testlib. -->
            <offlineLink>