[grpc-io] Java pattern for response observer that requires access to request observer?

I have a bi-directional RPC, let's use the sample StreamingGreeter . I need to register a response observer: // a response observer StreamObserver responseObserver = new StreamObserver() { ... };

Re: [grpc-io] Java pattern for response observer that requires access to request observer?

In the past, I created a final AtomicReference and assigned the requestObserver to it and then called a get on the atomic reference in the onNext of the response onserver when I wanted to ping pong back and forth. It works, but also felt a bit off, so if anyone has better I’d love to hear it!

[grpc-io] Re: Using DNS Authority in C# client

Just because it's related, I have an open issue (Name resolver support for C# missing ) which also has all those links in it and which I use for keeping this topic together and up2date. So you could subscribe to it to be notified when there is a good

[grpc-io] Re: Using DNS Authority in C# client

I just forgot to write the main idea I wanted to write... Sorry, still a bit tired... For the time until c-ares is merged, you could use the version from my PR and doing a simple GRPCLB implementation in C# that is just interacting with the HTTP interface from consul. Just to get things going

[grpc-io] Re: Using DNS Authority in C# client

AFAIK, c-ares should work on linux using the env flag, but for windows it still needs the PR you mentioned to get merged. Your best option would be to add a custom resolver in C++ which is then also available in C#. I run into the same problem and tried to solve it on a similar level, by adding

[grpc-io] Re: Using DNS Authority in C# client

Thanks for these posts. I think I understand the direction you are pointing me in. I will have a play and see if I can get it working with your PR. Any word from the team on when the c-ares resolver will be merged? Or is that the sosrt of thing I ought to ask in the mmunity meeeting? On

[grpc-io] Digest Authentication with gRPC

Hi , Does gRPC support digest authentication ? I am trying to figure out if that will work with the BigTable HBase client and I was told that it uses gRPC. Please let me know. -Shubhro -- You received this message because you are subscribed to the Google Groups "grpc.io" group. To

[grpc-io] Re: cannot use Handler ... as type grpc.methodHandler in field value

What was the protoc command you used to generate the pb.go file? It should look something like this: protoc --go_out=plugins=grpc:. *.proto If that doesn't work, I may need to see something more concrete (e.g. your actual .pb.go file; names redacted if necessary) to help. Can you file an

Re: [grpc-io] libuv, C++, gRPC

In principle, I think this should work, but it will probably be a little awkward because the libuv code was written to work with Node, and specifically with our own Node extension. So, there are a few caveats: 1. The Makefile isn't set up to build the libuv code, so you'll have to set

Re: [grpc-io] Updated RPC signature of service

On Mon, Dec 4, 2017 at 6:42 AM, Robert Bielik wrote: > Is there a way I can on the client side (C++ btw) check to see which > signature of the RPC that the server implements ? I imagine that > if I call the RPC with the wrong signature, I would get a >

Re: [grpc-io] Consequences for increasing flowControlWindow?

I'd suggest increasing the number only if you can show the result of (bandwidth * RTT) is higher than the current value (1 MB). We'd really like to be auto-tuning this value, but because we use such a high default value it's not been a huge burden. Go and C actually already auto-scale. And Java

Re: [grpc-io] What mode has the highest throughput?

On Sat, Dec 2, 2017 at 3:49 PM, wrote: > I guess this test is 'async streaming' according to the test name. But, > what streaming? Is it server-side, client-side or bidirectional? > It's streaming in "the one direction that matters." Since it is testing throughput in only

[grpc-io] Re: Custom Transport Security

Ya I was looking through the C core a bit last night and following through how SSL works. It definitely looks possible. That seems reasonable, are there any future plans to make such an API stable? Follow up question, can you have security (encryption/decryption) enabled/disabled at the RPC