alephzero

@rw1nkler you should be able to display just the `@file` description with ```rst .. doxygenfile:: nutshell.h :sections: detaileddescription ```

I've been building and testing on standard desktop linux. I'm happy to try out ARM / aarch64. No idea how easy or hard it will be to incorporate. Can you...

I created an AWS instance of `ubuntu/images/hvm-ssd/ubuntu-focal-20.04-arm64-server-20220131` The C++ parts of AlephZero builds and all the tests pass. I'm having a few problems with the Python version at the moment....

There are fuzz tests for multiple Publishers on the same topic, that get sigkilled at random times. That scenario has also been a common use case in production for a...

@pi1ot I added a fuzz test that I _think_ matches the scenario you described. Start 100s of processes that just publish continuously. Send a sigkill to each. Check that I...

I was able to run your code, but it didn't crash for me, though I only tried it with the master branch. I'll try it with v0.3 soon. Peaking at...

As an aside, in the current master, `a0::INIT_AWAIT_NEW` and `a0::ITER_NEXT` are optional defaults. You can now do: ```cpp s = new a0::Subscriber(t, [&i](a0::Packet pkt) { ```

Right now, rpc are built on pubsub and are very simple. RpcServer and RpcClient both have Publishers and Subscriber on the topic. An RpcClient publishes the request. The RpcServer callback...

You can do the following to avoid copying message, allocating a string, ... ```cpp p.pub(std::string_view(&m, sizeof(message))); ``` If you're pre-C++17, you can use the a0 drop in replacement: ```cpp p.pub(a0::string_view(&m,...

@ahao1995 The reader-writer mutex has some overhead compared to a raw mutex (a few additional atomic operations). Given that (in the zero-copy case) the lock is held long enough to...