Eric Anderson

The gradle configuration expects JDK 16, and it is using gradle's newer toolchain support: https://github.com/perfmark/perfmark/blob/0da8ae3d564fd8686f6f27a1ec8c3e217f31ec8c/build.gradle#L21-L25 The gradle toolchain will search for a working javac that supports Java 16 and [if...

Note for others: this is related to #8610 and its fix #8650. Hmm... that's troublesome. We can make it a compileOnly dependency, and maybe that will help, but I doubt...

Removing the dependency on OkHttp from OkHttpClientTransport and then making okhttp2 a compileOnly dependency sounds great. I had intended to do that, but haven't had time. And adding okhttp3 API...

@pouriaHemmati, are you seeing it when communicating with a C# server as well? Can you provide the additional debug data we had requested in https://github.com/grpc/grpc-java/issues/8797#issuecomment-1005907432 ?

I wonder if this is related to #8866. One is client-side and one is server-side, but the code I was poking through for it is mostly used by both sides...

Given retries were enabled in 1.40, it seems likely that there's an interaction with retries that are retaining data. Given that much data, the RetriableStream should commit itself. Maybe the...

@dapengzhang0, isn't the per_rpc_buffer_limit default in Java 1 MB though? The cross-RPC default limit is 16 MB. It seems they have at least 30 MB available, so I don't see...

I think it'd make sense to add a mode to the inprocess transport to have it serialize payloads so that the sizes are reported. I also wonder if we should...

> grpc/examples/route_guide.proto Darn. I missed that in #80. Neither of these example protos are in the correct directory. They should be in a directory that matches their name (so "helloworld"...

> Sanity tests will be added to verify that common proto files match the "ground truth" files contained here. I see the "ground truth" there as that they match at...