* Refactor end2end tests to exercise each EventEngine
* fix incorrect bazel_only exclusions
* Automated change: Fix sanity tests
* microbenchmark fix
* sanitize, fix iOS flub
* Automated change: Fix sanity tests
* iOS fix
* reviewer feedback
* first pass at excluding EventEngine test expansion
Also caught a few cases where we should not test pollers, but should
test all engines. And two cases where we likely shouldn't be testing
either product.
* end2end fuzzers to be fuzzed differently via EventEngine.
* sanitize
* reviewer feedback
* remove misleading comment
* reviewer feedback: comments
* EE test_init needs to play with our build system
* fix golden file test runner
Co-authored-by: drfloob <drfloob@users.noreply.github.com>
* eliminate unnecessary templates
* add xDS node ID to error messages generated by XdsClient
* change test to verify that node ID is present in status message
* clang-format
This commit update the scenario results to include the start_timestamp
and end_timestamp. These two fields are used in prometheus range queries
to indicate the period we are pulling data for.
* Make grpc_core::Call an interface type
Initially this won't go down in the chronicals of systems design as one
of the finest types ever designed. It exists right now to allow us to
begin to pivot from a filter stack based call structure to promises.
We'll iterate on this type and its interface until promises are stood
up, and then we'll drop this abstraction and return to a direct
implementation against a (new) concrete type.
* fix some bad casts
* fix deadlines
* wtf
* Add bazel cpp distribtest for grpc_cc_library
* Add new directory to bazelignore
* Try changing to workspace differently
* Debug cd
* Fix bazel wrapper
* Yet more debug
* Try again
* Clean up
* Sanity fixes
* Run per-language distribtests even if overall build fails
* priority LB: avoid possibility of rescheduling a timer before it fires
* clang-format
* fix memory leak
* small change, just to be paranoid
* inline StartFailoverTimerLocked()
* initialize timer_pending_ to true
* don't check shutting_down_ in timer callbacks
* xds_cluster_resolver: fix bug that caused us to incorrectly reuse priority child names
* add debug logging
* assign child policy names within each discovery mechanism, not across all mechanisms
* clang-format
* add tests
* don't need EDS service name or LOGICAL_DNS hostname in child name
* remove unnecessary code
* client channel: record call completion even if recv_trailing_metadata was not started
* add test
* add test for opencensus filter
* remove unnecessary code
NdkBinder expects the thread to be already attached to JVM when
AIBinder_transact is called with a local Java binder. If that is not the
case, a null pointer dereference will happen in NdkBinder.
This commit tries to cache the pointer to JVM when a new connection is
created (one of the from/toJavaBinder will be called in that case) and
tries to ensure the JVM is attached to current thread before calling
AIBinder_transact.
Yash Tibrewal
Craig Tiller
Richard Belleville
apolcyn
Alisha Nanda
donnadionne
Nicolas Noble
Mark D. Roth
AJ Heller
Jan Tattermusch
Wanlin Du
tomerv
Lidi Zheng
yihuaz
Mahdi K. Fard
yifeizhuang
krestofur
Ming-Chuan