Previously, it always threw a RuntimeExecutionException with a
StatusRuntimeException within (since the only callers of setException()
provide a StatusRuntimeException).
Resolves#507
On my machine, the client currently takes ~3s to run a test. However,
with exit() it takes < 1s. When doing lots of integration tests, those
seconds add up.
We know one of those seconds is the DESTROY_DELAY_SECONDS of
SharedResourceHolder. Part of another second appears to be Netty
creating any threads that weren't previously created when shutting down.
1. Move DEFAULT_CONNECTION_SPEC to OkHttpChannelBuilder
2. make OkHttpClientTransport package-private
3. Rename OkHttpChannelBuilder.setConnectionSpec to connectionSpec
GCM is very slow, and doesn't provide any benefit in unit tests. Even if
we were using tcnative and GCM is fast, using more available ciphers in
tests still makes sense. With this change building with Java 7 works
again, although that isn't the reason for the change.
On my machine with parallel building, it cuts full build time from
92 seconds to 39 seconds. For an incremental build after only changing
an interop test, the build time is cut from 73 seconds to 15 seconds.
Other classes are already following the convention that ClientFoo for
client-side, and ServerFoo for server-side. Call has been the black
sheep of the family.
- Call -> ClientCall
- Calls -> ClientCalls
- ForwardingCall* -> ForwardingClientCall*
To use the test certs a module has to depend on the grpc-interop-testing module, which doesn't really make sense. I'm moving the certs to the grpc-testing module, which is meant to be a sort of testing common area for all of our modules.
We'll probably change our recommendations later to suggesting
netty-tcnative or maybe even conscrypt, but for the moment simply
documenting the issue is productive to prevent surprises.
The classes are available, even on Windows. Trying to use them though
won't work. You'll get an error like:
java.lang.UnsatisfiedLinkError: no netty-transport-native-epoll in java.library.path
This doesn't catch 100% of such problems, as it runs before tests have
been run. However, almost all of our protos are used in 'main' so this
limitation shouldn't be a big deal.
If it does become a problem, we change it from 'before_script' to
'script', but then we'll also need to include gradle build instructions.
"Interoperability" is a more appropriate name for the tests, since they
are used for testing across different implementations. They will do a
bit of integration testing, like for auth, but this is a smaller scale.
It seems the other languages (Go, C++, Node, PHP, Python, Ruby) are
using "interop" to describe the tests, and the test case specifications
document is named with "interop". After this change, C# will be the only
language calling them "integration" tests.
This change just renames the folder and artifact. We can change the
internal package names later. However, once we do a release, old
artifact names will live forever in Maven Central.
Eric Anderson
Xudong Ma
Jack Coughlin
nmittler
David Connelly
Louis Ryan
Josh Humphries
Kun Zhang
zsurocking
jonathanlin