Merge lp://qastaging/~mir-team/mir/port-uncooperative-tests-to-surface-spec into lp://qastaging/mir

I'm not sure that's the right fix? Maybe it's the best we can do given protobuf's callbacks?

19 - send_message(invocation, invocation, fds);
20 + try
21 + {
22 + send_message(invocation, invocation, fds);
23 + }
24 + // In general we propagate this exception up to the client
25 + // library so that it may invoke appropriate error callbacks,
26 + // etc...however in the case of disconnect the notify_disconnect
27 + // codepath will take care of this and we need to surpress the
28 + // exception here. (lp:1394873)
29 + catch (...)
30 + {
31 + if (!disconnected.load())
32 + throw;
33 + }

This is not really anything to do with the stated reason for this MP is it?

>> his is not really anything to do with the stated reason for this MP is it?

https://code.launchpad.net/bugs/1394873

Is it the branch title you'd like me to change?

So, fix + port. Probably wanna separate the fix out in its own MP.

For the ports, same comments apply :
https://code.launchpad.net/~mir-team/mir/port-cooperative-tests-to-surface-spec/+merge/244056/comments/601848

Approving this. Port is ok (as mentioned in the other MP).

> >> his is not really anything to do with the stated reason for this MP is it?
>
> https://code.launchpad.net/bugs/1394873
>
> Is it the branch title you'd like me to change?

No, just confirming my understanding

$ bin/mir_unit_tests --gtest_filter=MirProtobufRpcChannelTest.*
Running main() from command_line_server_configuration.cpp
Note: Google Test filter = MirProtobufRpcChannelTest.*
[==========] Running 8 tests from 1 test case.
[----------] Global test environment set-up.
[----------] 8 tests from MirProtobufRpcChannelTest
[ RUN ] MirProtobufRpcChannelTest.ReadsFullMessages
[ OK ] MirProtobufRpcChannelTest.ReadsFullMessages (0 ms)
[ RUN ] MirProtobufRpcChannelTest.ReadsAllQueuedMessages
[ OK ] MirProtobufRpcChannelTest.ReadsAllQueuedMessages (0 ms)
[ RUN ] MirProtobufRpcChannelTest.SendsMessagesAtomically
[ OK ] MirProtobufRpcChannelTest.SendsMessagesAtomically (1 ms)
[ RUN ] MirProtobufRpcChannelTest.SetsCorrectSizeWhenSendingMessage
[ OK ] MirProtobufRpcChannelTest.SetsCorrectSizeWhenSendingMessage (0 ms)
[ RUN ] MirProtobufRpcChannelTest.ReadsFds
[ OK ] MirProtobufRpcChannelTest.ReadsFds (1 ms)
[ RUN ] MirProtobufRpcChannelTest.NotifiesOfDisconnectOnWriteError
/home/alan/display_server/mir4/tests/unit-tests/client/test_protobuf_rpc_channel.cpp:321: Failure
Expected: channel_user.exchange_buffer(nullptr, &request, &reply, google::protobuf::NewCallback([](){})) throws an exception of type std::runtime_error.
  Actual: it throws nothing.
[ FAILED ] MirProtobufRpcChannelTest.NotifiesOfDisconnectOnWriteError (0 ms)
[ RUN ] MirProtobufRpcChannelTest.ForwardsDisconnectNotification
[ OK ] MirProtobufRpcChannelTest.ForwardsDisconnectNotification (0 ms)
[ RUN ] MirProtobufRpcChannelTest.NotifiesOfDisconnectOnlyOnce
/home/alan/display_server/mir4/tests/unit-tests/client/test_protobuf_rpc_channel.cpp:382: Failure
Expected: channel_user.exchange_buffer(nullptr, &request, &reply, google::protobuf::NewCallback([](){})) throws an exception of type std::runtime_error.
  Actual: it throws nothing.
[ FAILED ] MirProtobufRpcChannelTest.NotifiesOfDisconnectOnlyOnce (1 ms)
[----------] 8 tests from MirProtobufRpcChannelTest (3 ms total)

[----------] Global test environment tear-down
[==========] 8 tests from 1 test case ran. (4 ms total)
[ PASSED ] 6 tests.
[ FAILED ] 2 tests, listed below:
[ FAILED ] MirProtobufRpcChannelTest.NotifiesOfDisconnectOnWriteError
[ FAILED ] MirProtobufRpcChannelTest.NotifiesOfDisconnectOnlyOnce

 2 FAILED TESTS

I think this is the right fix now...(in r2133).

In the case that send_message throws, the MirSurface constructor will now catch it and convert it in to an error. In this case the client library wont catch an exception and wont construct the duplicate error object.

The exception catch + error object creation is left in for cases where MirSurface constructor really does throw (e.g. as in the test which simulates BufferFactory throw failure).

It was necessary to make some changes in MirSurface::created to handle the pending call completion for the surface which never received a response.

FAILED: Continuous integration, rev:2133
http://jenkins.qa.ubuntu.com/job/mir-ci/2338/
Executed test runs:
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-android-vivid-i386-build/445
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-clang-vivid-amd64-build/445
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-vivid-touch/419/console
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-vivid-amd64-ci/335/console
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-builder-vivid-armhf/419/console

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/2338/rebuild

OK

FAILED: Continuous integration, rev:2135
http://jenkins.qa.ubuntu.com/job/mir-ci/2401/
Executed test runs:
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-android-vivid-i386-build/512
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-clang-vivid-amd64-build/512
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-vivid-touch/486/console
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-vivid-amd64-ci/398/console
    SUCCESS: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-builder-vivid-armhf/486
        deb: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-builder-vivid-armhf/486/artifact/work/output/*zip*/output.zip
    FAILURE: http://jenkins.qa.ubuntu.com/job/mir-mediumtests-runner-mako/3669/console
    SUCCESS: http://s-jenkins.ubuntu-ci:8080/job/touch-flash-device/16562

Click here to trigger a rebuild:
http://s-jenkins.ubuntu-ci:8080/job/mir-ci/2401/rebuild