The proxying code was responding to a `Close` message by ending the pump loop, which had the effect of immediately dropping the connection after a `Close` had been seen in each direction. This is arguably incorrect behavior: for proxying purposes, `Close` messages and underlying TCP disconnects should be treated as independent events, forwarded separately.

In practice this "bug" probably would never cause a problem and perhaps doesn't even violate spec (since `Close` was seen in both directions). But, OSX's implementation of shutdown() returns ENOTCONN if the connection has already been disconnected from the remote end. This is the case here, as the proxy dropped all connections immediately after sending the final `Close`.

This in turn led to a unit test failure.

The intended behavior was that the proxy would forward exactly what it saw: If a `Close` was sent, it would be forwarded, without changing the underlying connection state. If a TCP disconnect was detected, it would be "forwarded" by disconnecting the next leg. This change implements that behavior.

Due to differences in the way I/O events are queued on Windows, the timing of this test end up different, such that the two incoming pings are not both received before the large outgoing message is sent.

To fix this, I removed the dependency on native I/O altogether by implementing an in-memory pipe that does no buffering (it requires a read() and write() to rendezvous, then copies between their buffers).

Still need to add handshake separately.

Up until now, I've always felt kj::OneOf had a rather janky interface, which made me avoid using it in other interfaces. With this macro I'm pretty happy with it as *the* way to define variants.

It turns out this is more than just android. Some Debian Stretch docker image I'm working with seems to have this problem (but Stretch in general does not??).

Fix link errors with -fsanitize=vptr

Add new `capnp convert` command with JSON support.

This supersedes the `capnp encode` and `capnp decode` commands. It didn't make sense to add JSON to those commands since it was unclear if JSON should be thought of as the "encoded" or "decoded" format. `convert` allows mapping anything to anything.

This command is also useful for, say, converting unpacked format to packed format or vice versa, which can now be done without a schema.

Allow an ExceptionCallback to control how new threads' ExceptionCallbacks are initialized.

Fix minor typo: varibale -> variable

Clarify RPC spec documentation based on mail thread

Kenton rightly pointed out that this is an unnecessary round-trip.

Update README.md

protobuf link update

Currently only implemented for Linux (when using futexes), since that's what I use. Will expand to other platforms later.

synchrozines behavior with autoconf build

https://github.com/capnproto/capnproto/issues/515

https://groups.google.com/d/topic/capnproto/_1kBtRSC51s/discussion

Doc: Enforce comment-after-code rule

Install capnp-json.pc