Conversation
…supported in v1.0.x This reverts commit 59af1bf.
This reverts commit 1079040.
Backport: Send RST_STREAM from client when it receives trailing metadata without the corresponding RST_STREAM
Re-enable Node 7 artifact build
Advance objective c version to v1.0.2
wait for write loop to finish at end of ruby read loop, on client side calls
The requirement that any created managed call must have operations performed on it is obstructing proper handling of the case of applications providing invalid invocation metadata. In such cases the RPC is "over before it starts" when the very first call to start_client_batch returns an error.
…ED-STEVE Backport of grpc#9071
Node: correctly bubble up errors caused by non-serializable writes
…ady_future_tests-v1.0.x Channel ready future test tweaks.
…xes-v1.0.x Interop fixes.
…-you-do-the-wrong-thing-will-it-still-do-the-right-thing Patch overlooked strings from Python un-namespacing
add ruby subclasses of bad status for each GPRC status code
Add check on return value from start_client_batch
| rescue GRPC::BadStatus => e | ||
| bad_status_code = e.code | ||
| rescue GRPC::ResourceExhausted | ||
| one_failed_as_unavailable = true |
There was a problem hiding this comment.
Something is definitely off with this merge. This line changes from bad_status_code to one_failed_as_unavailable, but the check on line 425 still looks at bad_status_code.
9453657 to
c8de2d2
Compare
| expected_error_details = "metadata was invalid: %s" % metadata | ||
| with self.assertRaises(ValueError) as exception_context: | ||
| self._unary_unary(request, metadata=metadata) | ||
| self.assertIn(expected_error_details, str(exception_context.exception)) |
There was a problem hiding this comment.
Other option would be to swap expected_error_details with expected_error of type ValueError, then compare the str() of those two. But this works equally as well and probably has less unneeded overhead. Looks good!
There was a problem hiding this comment.
Thanks! I want to keep it in this form because I don't want to constrain the system under test to always raise a ValueError that always has a certain exact string. This test condition comes close to that, sure, but it at least buys a little flexibility for the system under test.
Manual changes: - Fixed use of Exception.message in _invalid_metadata_test.py - Fixed merge of one_failed_as_unavailable in rpc_server_spec.rb - Added "set -e" to generate_build_additions.sh
c8de2d2 to
8f7b0b4
Compare
|
LGTM on ruby changes |
|
@nicolasnoble: tests are all green; please review and merge. 😄 |
|
That looks good to me overall. |
No description provided.