Eric Gribkoff

Can you elaborate on why `getState` and `notifyStateChanged` are unsuited for your use case, and what makes them unreliable versus the proposed error callback? Your proposed error callback also has...

I sent out some PRs to clarify the documentation. We can use this issue to track demand for on-device servers. I'll note that we intend to support gRPC C++ servers...

There is no maven artifact that should be used instead. The proto was marked as "deprecated" in https://github.com/grpc/grpc-java/pull/4973, but AFAICT there has been no effort since then to move any...

Failed at https://grpc-testing.appspot.com/job/gRPC-Java-PR-Windows/3316/console ``` io.grpc.testing.integration.NettyFlowControlTest > verySmallWindowMakesProgress FAILED java.lang.AssertionError: Window was 687918 expecting 314570 at org.junit.Assert.fail(Assert.java:88) at org.junit.Assert.assertTrue(Assert.java:41) at io.grpc.testing.integration.NettyFlowControlTest.doTest(NettyFlowControlTest.java:169) at io.grpc.testing.integration.NettyFlowControlTest.verySmallWindowMakesProgress(NettyFlowControlTest.java:139) ```

https://kokoro.corp.google.com/job/grpc/job/java/job/master/job/windows/job/presubmit/job/windows/157/ ``` java.lang.AssertionError: Window was 720686 expecting 314570 at org.junit.Assert.fail(Assert.java:88) at org.junit.Assert.assertTrue(Assert.java:41) at io.grpc.testing.integration.NettyFlowControlTest.doTest(NettyFlowControlTest.java:154) at io.grpc.testing.integration.NettyFlowControlTest.verySmallWindowMakesProgress(NettyFlowControlTest.java:124) ```

https://kokoro.corp.google.com/job/grpc/job/java/job/master/job/windows/job/presubmit/job/windows/204/ ``` java.lang.AssertionError: Window was 655360 expecting 314570 at org.junit.Assert.fail(Assert.java:88) at org.junit.Assert.assertTrue(Assert.java:41) at io.grpc.testing.integration.NettyFlowControlTest.doTest(NettyFlowControlTest.java:154) at io.grpc.testing.integration.NettyFlowControlTest.verySmallWindowMakesProgress(NettyFlowControlTest.java:124) ```

Spoke with @groakley and he mentioned that @wanyingd1996 might be interested in taking this on. But apparently I can't directly assign them... (previous contributor in https://github.com/grpc/grpc-java/pull/7242)

> At a minimum, we'd want to make sure that any client-implemented method that throws doesn't hang the stub or prevent resources from being released. Yes. We also will want...