Convert names_and_types graph APIs to pybind11#717
Conversation
gbalke
left a comment
There was a problem hiding this comment.
Looks good! Couple comments/questions.
rclpy/src/rclpy/_rclpy_pybind11.cpp
Outdated
| m.def( | ||
| "rclpy_get_service_names_and_types", | ||
| &rclpy::graph_get_service_names_and_types, | ||
| "Get all service names and types in the ROS graph."); | ||
| m.def( | ||
| "rclpy_get_service_names_and_types", | ||
| &rclpy::graph_get_service_names_and_types, | ||
| "Get all service names and types in the ROS graph."); |
| * Raises ValueError if pywait_set is not a wait set capsule | ||
| * Raises RuntimeError if the entity type is not known | ||
| * |
There was a problem hiding this comment.
Where is pywait_set coming from?
sloretz
left a comment
There was a problem hiding this comment.
It looks like the removal of code from _rclpy.c is missing from the PR, mind adding it?
Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
4c66f3d to
f009f08
Compare
|
Rebased to solve conflicts. |
sloretz
left a comment
There was a problem hiding this comment.
One nitpick that I wouldn't block on.
Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
|
@hidmic, It looks like this one is ready to go! I'll merge it so it doesn't need to be rebased again if something else would have gotten merged first. |
|
I'm not 100% sure if it was this PR, but Windows debug build is now broken in rclpy: https://ci.ros2.org/view/nightly/job/nightly_win_deb/1931/ . The only two PRs that were merged in rclpy yesterday were this one and #712 . @sloretz @hidmic can you take a look? |
|
It's trying to link to non-debug Python libraries: but the CMakeLists.txt does seem to be taking care of not doing that. Has this happened to you before @sloretz ? |
Yeah, this seems familiar. There's some odd linker pragmas in the python.h and pybind11.h headers involving the |
|
I think we've been getting lucky with the order. I'm guessing the linker pragma set by the last compilation unit is the one that wins :-/ The jobs that passed, the C++ file comes last: The job that failed, a C file came last |
|
Oh my. Will fix now. |
Precisely what the title says. Part of #665. This patch also brings
rcpputilsas a dependency.CI up to
rclpy: