fix .travis.rosinstal to use fkanehiro/hrpsys-base #248
Conversation
There was a problem hiding this comment.
Nice! I overlooked this issue.
Please also update openhrp3 like https://github.com/start-jsk/rtmros_common/blob/master/.travis.rosinstall.kinetic#L1-L4, because https://github.com/start-jsk/openhrp3 also seems deprecated.
In addition, could you comment out the following allow_failures:
Lines 35 to 38 in 26a9dd2
like https://github.com/jsk-ros-pkg/jsk_pr2eus/blob/master/.travis.yml#L34-L35?
This PR may fix failure of those
USE_DEB=false tests.
|
It takes too long time to test with I found a bug on using |
Naoki-Hiraoka
force-pushed
the
fix-travis.rosinstal
branch
from
7156bca
to
d4a0c24
Compare
|
I changed only |
| - git: | ||
| uri: https://github.com/start-jsk/rtmros_tutorials | ||
| local-name: rtm-ros-robotics/rtmros_tutorials | ||
| # rtmros_tutorials depends on rtmros_gazebo. So rtmros_tutorials is not included here. |
There was a problem hiding this comment.
rtmros_tutorials is added at 2a8b8a4.
And I think rtmros_tutorials should be here, because when we want to change base packages in rtmros_gazebo (e.g., hrpsys_gazebo_general), we have to check this with rtmros_tutorials, which is what we use.
rtmros_common has tests from this thought: https://github.com/start-jsk/rtmros_common/blob/master/.travis.yml#L31
There was a problem hiding this comment.
Oops, I don't know that rtmros_gazebo/hrpsys_gazebo_atlas depends on rtmros_tutorials/hrpsys_ros_bridge_tutorials...
My previous comment is still effective, but I also think this dependency should be resolved.
There was a problem hiding this comment.
I think when we want to change base packages in rtmros_gazebo (e.g., hrpsys_gazebo_general), we only have to check this with SampleRobot, which can be simulated without rtmros_tutorials ???
There was a problem hiding this comment.
I see. I forgot about SampleRobot.
And I also forgot that HRP2 is closed robot and cannot be tested in this repository.
So problem of removing rtmros_tutorials seems mainly about hrpsys_gazebo_atlas.
You are preventing hrpsys_gazebo_atlas from being built, aren't you?
I'm against this idea, because we will hardly notice when we break
hrpsys_gazebo_atlas until we explicitly check travis of rtmros_tutorials.
I think this makes travis meaningless.
There was a problem hiding this comment.
How about leaving rtmros_tutorials for now, and fixing hrpsys_gazebo_atlas not to depend on rtmros_tutorials in anorther PR? (that PR can be postponed)
Or that solution cannot work on melodic?
There was a problem hiding this comment.
You are preventing hrpsys_gazebo_atlas from being built, aren't you?
In this commit, hrpsys_gazebo_atlas is built, but not tested.
In this commit,
BUILD_PKGS="",
TEST_PKGS="eusgazebo hrp2jsk_moveit_config hrp2jsknt_moveit_config hrp2jsknts_moveit_config hrp2w_moveit_config hrpsys_gazebo_msgs hrpsys_gazebo_general samplerobot_moveit_config staro_moveit_config".
In travis.sh,
catkin build $BUILD_PKGS #build all packages
catkin build --catkin-make-args run_tests -- -i --no-deps --no-status $TEST_PKGS #not test hrpsys_gazebo_atlas
fixing hrpsys_gazebo_atlas not to depend on rtmros_tutorials
Naoki-Hiraoka
force-pushed
the
fix-travis.rosinstal
branch
from
78cbeb5
to
5eb1b94
Compare
|
This PR is included in #249 |
Currently
rtmros_gazebois tested with https://github.com/start-jsk/hrpsys, which is a deprecated repository..travis.rosinstalis fixed to use https://github.com/fkanehiro/hrpsys-base.