Testing xenial build on Travis (DO NOT MERGE) #279
Closed
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…ing a temporary object Signed-off-by: Johannes Meyer <[email protected]>
The previous implementation based on std::rand() used time() to seed the random number generator on every call. If multiple connections are generated within the same second, they can end up with the same ID. UUIDs according to IETF RFC 4122 have a very high probably to be unique. Signed-off-by: Johannes Meyer <[email protected]>
The buffer_policy field will be composed from and decomposed to int because CPFMarshaller cannot handle enum BufferPolicy as a primitive type. Signed-off-by: Johannes Meyer <[email protected]>
Moved portqueue from class TaskContext to the ExecutionEngine Reviewed and this is the desired behavior.
…-queue' into toolchain-2.9 Conflicts: tests/ports_test.cpp
…p' into toolchain-2.9
This line was missing from commit d9ff6ab. For details, see #91. Signed-off-by: Johannes Meyer <[email protected]>
Signed-off-by: Johannes Meyer <[email protected]>
…dataflow_performance_test For the case the user running the tests has no permissions to use the RT scheduler, the test should not fail completely. However, quantitative results may depend largely on system load. The test runner on Travis failed because of this issue, e.g. https://travis-ci.org/orocos-toolchain/rtt/builds/105478386. Signed-off-by: Johannes Meyer <[email protected]>
…into toolchain-2.9
…dering' into toolchain-2.9
…ad' into toolchain-2.9
This reverts commit 49fe051. The original implementation of #109 required #91 (more verbose ActivityInterface and RunnableInterface), which was not yet merged into master at that time. This commit reverts the original commit which adapted the PR to be compatible with master. Signed-off-by: Johannes Meyer <[email protected]>
… without a runner ...and extended the task test to cover this bug and to test non-periodic activities. Signed-off-by: Johannes Meyer <[email protected]>
Call step() from Activity::loop() for the non-periodic case without a runner
If entry is not done yet, SM would immediately transition to next state instead of finishing entry program. - An event during entry will fail if entry could not be executed atomically (in a single SM step() ). - We're now transitioning to the other state while processing the event callback in eventTransition() ( stepping=false) Unit tests: - Don't read same variable in the global event transitions as in local state transitions, which queries the port and fills it in (see testStateEvents ). - The testStateOperationSignalGuard is not fixed yet, it receives the last event value instead of the first, while both events do arrive in the event queue. It appears that though the callbacks are asyn processed, the datasource is already written in both cases, so the first callback executed already sees the value of the second invocation of 'mo'. Signed-off-by: Peter Soetens <[email protected]> Conflicts: rtt/scripting/StateMachine.cpp tests/state_test.cpp
…nstructor to ready() member function ... so that it can be reused in corba_ipc_server without having initialized the Boost Test framework. This is a regression from 167076a. Signed-off-by: Johannes Meyer <[email protected]>
…llingFromSlave test This patch should have been included in the merge commit 7643669. Signed-off-by: Johannes Meyer <[email protected]>
…tantiation The loop that removed SendHandleAlias variables from the StateMachineService could have skipped the second one for the case the first two variables are SendHandles. Signed-off-by: Johannes Meyer <[email protected]>
... that was introduced in 3b29823. Signed-off-by: Johannes Meyer <[email protected]>
This section in rtt/CMakeLists.txt was forgotten to be removed in 079dbb3. Signed-off-by: Johannes Meyer <[email protected]>
Signed-off-by: Johannes Meyer <[email protected]>
…stAffinity The condition to execute that part was always evaluating to false because it compared a scheduler type with a priority. The actual test case is invalid because even for a real-time thread with the highest priority there is no guarantee that it executes at least once between the start and stop call. Signed-off-by: Johannes Meyer <[email protected]>
collectIfDone(...) was calling cimpl->collect(...) for the specializations with more than three arguments. Also reordered template specializations in Collect.hpp. Signed-off-by: Johannes Meyer <[email protected]>
This reverts commit 2318f82.
refactor Thread*::getCpuAffinity() to avoid changing the vtable. keep unused argument of ComponentLoader::unloadComponent(...).
…2.9 (#271) * Update README * Rename README to README.md
boost::serialization::array<T> has been renamed to array_wrapper<T> in Boost 1.61. Its usage in RTT has been adapted in 8765918 with a follow-up in 0d83c90. The respective header has been renamed to array_wrapper.hpp in Boost 1.64 and starting from 1.65 array.hpp includes array_wrapper.hpp (boostorg/serialization@1d86261). But for 1.64 compilation failed with the following errors: rtt/internal/../types/carray.hpp:101:56: error: expected ‘)’ before ‘<’ token carray( boost::serialization::array_wrapper<T> const& orig) rtt/internal/../types/carray.hpp:171:41: error: ‘array_wrapper’ is not a member of ‘boost::serialization’ const carray<T>& operator=( boost::serialization::array_wrapper<OtherT> const& orig ) { ^ rtt/internal/../types/carray.hpp:171:83: error: expected primary-expression before ‘>’ token const carray<T>& operator=( boost::serialization::array_wrapper<OtherT> const& orig ) { ^ rtt/internal/../types/carray.hpp:171:85: error: expected primary-expression before ‘const’ const carray<T>& operator=( boost::serialization::array_wrapper<OtherT> const& orig ) { ^ rtt/internal/../types/carray.hpp:171:97: error: declaration of ‘operator=’ as non-function const carray<T>& operator=( boost::serialization::array_wrapper<OtherT> const& orig ) { ^ Signed-off-by: Johannes Meyer <[email protected]>
It depends on "normal" timing and tends to fail on Travis CI. See also 7edd7ed. Signed-off-by: Johannes Meyer <[email protected]>
Signed-off-by: Johannes Meyer <[email protected]>
Signed-off-by: Johannes Meyer <[email protected]>
…aces from API documentation Signed-off-by: Johannes Meyer <[email protected]>
…of the RTT cmake config If the RTT cmake module is used more than once with different components and the REQUIRED argument was not given during the first invocation, a second invocation with REQUIRED might trigger false errors because of items added to MISSING_COMPONENTS earlier. Also moved initialization of OROCOS-RTT_PLUGIN_PATH to the if(NOT OROCOS-RTT_FOUND) section to not readd the same directories over and over again. Signed-off-by: Johannes Meyer <[email protected]>
This fixes a regression from 75745d6: The ListLockFree instance was instantiated with a maximum number of threads of 4, but the test case ran 5 workers and one additional grower threads. The number of workers has now been reduced to 3 again, like before 75745d6. The fixture class was actually not used for the last test cases. Only the `listlockfree` member, which has been moved into the `testListLockFree()` test case. This bug caused occassional build errors on Travis CI, e.g. https://travis-ci.org/orocos-toolchain/rtt/builds/518819992. Signed-off-by: Johannes Meyer <[email protected]>
|
Please rebase this patch on See also the related PR #195, which was an attempt to switch to containerized builds and also compile with different compilers. Unfortunately this build infrastructure has been (deprecated again in the meantime](https://docs.travis-ci.com/user/reference/trusty/#container-based-infrastructure). I think the best option would be to follow-up in #195 and do the switch to |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
PR for testing xenial build on Travis. Once we have the .yml correct, we can cherry pick .travis.yml to
masterand both release branches.