Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    Wolf Vollprecht
    @wolfv
    cool, yeah, the most common choices are either the "provision-with-micromamba" action, or the https://github.com/conda-incubator/setup-miniconda action
    I am biased towards the former micromamba one :)
    Michael Carroll
    @mjcarroll
    Perfect, I will give that a look over
    1 reply
    Andrés Morales
    @mora200217
    Hi, im trying to access a function from a different file but I keep getting the symbol not found.
    both files are in the same ros package
    I'm not sure what to do
    Tobias Fischer
    @Tobias-Fischer
    Could you please provide more information (what's the build error, which package, is it specific to RoboStack, link to code
    etc.)?
    Wolf Vollprecht
    @wolfv
    @mora200217 if you have "symbol not found" it's usually that you're not linking properly to the other library
    Andrés Morales
    @mora200217
    yup, it was a dummie mistake. In the CmakeList I dind't link correctly the library .
    Silvio Traversaro
    @traversaro:matrix.org
    [m]
    On other news, I saw in the ROSCon presentations (icub-tech-iit/tickets#2272) one on the ROS-O initiative, that is apparently is a project to continue maintaining ROS1 packages, that may be interesting for RoboStack's ros-noetic packages
    1 reply
    Wolf Vollprecht
    @wolfv
    Indeed. I actually had a call with Michael
    I wanted to update you all on it ... I think it's a cool initiative!
    Silvio Traversaro
    @traversaro:matrix.org
    [m]
    great!
    Silvio Traversaro
    @traversaro:matrix.org
    [m]

    Yeah, it would be nice to reduce the amount of patching that is going on here

    I was hoping that to happe in ogre-next 2.4/3.0, see OGRECave/ogre-next#232 and OGRECave/ogre-next#259, but I am not sure when that is actually going go happen

    however, indeed the vulkan-loader may not be the simplest package to start from,due to the interaction with X11/Wayland/etc/etc
    Michael Carroll
    @mjcarroll
    I think it's largely just a mismatch between support windows and API/ABI philosophy
    We can get some of it cleaned up by making the way we package with conda/deb consistent
    Tobias Fischer
    @Tobias-Fischer
    I fixed the qt-gui-cpp issues (at least on linux-64), all it needed was the following: RoboStack/vinca@43280d5
    Tobias Fischer
    @Tobias-Fischer
    Do you guys know of a generic way to obtain the platform specifier (x86_64-linux-gnu / darwin / aarch) etc.?
    (In cmake)
    Wolf Vollprecht
    @wolfv
    nice!
    hmm, I guess uname might work on Unix (not Win though)
    Maybe the variables in "Variables that describe the system" https://cmake.org/cmake/help/v3.0/manual/cmake-variables.7.html#id4
    Michael Carroll
    @mjcarroll
    @Tobias-Fischer what specifically do you need it for?
    There are a whole bunch of ways to do anything in CMake, just trying to understand the application to narrow the scope a bit
    Silvio Traversaro
    @traversaro:matrix.org
    [m]
    Indeed, those seems different things x86_64-linux-gnu seems a gnu triplet (i.e. what you get from gcc -dumpmachine), darwin an os (so the last element of a GNU triplet), aarch an architecture (first element of a triplet)
    Tobias Fischer
    @Tobias-Fischer
    I think I managed it .. thanks all :)
    It's for a patch for qt-gui-cpp where the library is named like that. In case of Linux it seems to be the gnu triplet, and in case of MacOS just darwin
    I have no idea how to patch qt-gui-cpp for Windows though - if someone has some capacity to setup a local build of ros-humble on Windows this would be awesome (let me know and I can provide more details)
    Tobias Fischer
    @Tobias-Fischer
    @wolfv: please have a look at conda-forge/cartographer-feedstock#43
    Tobias Fischer
    @Tobias-Fischer
    Also @wolfv - any idea what's going on with the aarch sever? Is there some limits in place? The self-hosted runner: cirun-robostack--ros-humble-bedf657 lost communication with the server. Verify the machine is running and has a healthy network connection. Anything in your workflow that terminates the runner process, starves it for CPU/Memory, or blocks its network access can cause this error.
    Tobias Fischer
    @Tobias-Fischer
    So I think the issue with qt-gui-cpp on osx is that it uses the system clang rather than the conda clang. It's weird as $BUILD_PREFIX/bin/clang already exists. Not sure where it picks up the wrong compiler .. do you think we should just remove it?
    Wolf Vollprecht
    @wolfv
    Hmm, that is weird, the linux-aarch64 machines are started from cirun
    So they should be fresh
    Silvio Traversaro
    @traversaro:matrix.org
    [m]

    I have no idea how to patch qt-gui-cpp for Windows though - if someone has some capacity to setup a local build of ros-humble on Windows this would be awesome (let me know and I can provide more details)

    Unfortunatly this week I am quite constrained and next week I will leave for vacations, so I guess I can do it around ~20 of August

    Tobias Fischer
    @Tobias-Fischer
    Awesome that would be great
    I'll go on parental leave any day now ;)
    Silvio Traversaro
    @traversaro:matrix.org
    [m]
    Congratulations!
    Tobias Fischer
    @Tobias-Fischer
    Thanks Silvio :)
    @wolfv: I think there's a 25 minute job limit for the linux-aarch builds .. could this be?
    Tobias Fischer
    @Tobias-Fischer
    linux-64 rebuild for humble with qt-main is complete ..
    Wolf Vollprecht
    @wolfv
    Hmm, strange, there should not be any job limit
    I've just asked on the cirun.io chat
    Dan Zimmerman
    @danzimmerman
    Should I be able to upgrade packages that need Qt 5.15 in a linux64 Humble env? I'm looking to install MoveIt Setup Assistant and getting
     mamba install ros-humble-moveit-setup-assistant --dry-run
    Encountered problems while solving:
      - package ros-humble-moveit-setup-assistant-2.5.3-py39h7ac24ad_1 requires qt-main >=5.15.4,<5.16.0a0, but none of the providers can be installed
    14 replies
    This message was deleted
    Dan Zimmerman
    @danzimmerman

    Looks like moveit_setup_assistant is assuming Python 3.10+ dataclasses (in moveit_configs_utils) with a slots kwarg in the @dataclass decorator.

    (humble-msa) dan@schelkunoff:~$ ros2 launch moveit_setup_assistant setup_assistant.launch.py debug:=true
    [INFO] [launch]: All log files can be found below /home/dan/.ros/log/2022-08-03-18-34-30-806882-schelkunoff-77251
    [INFO] [launch]: Default logging verbosity is set to INFO
    [ERROR] [launch]: Caught exception in launch (see debug for traceback): Caught exception when trying to load file of format [py]: dataclass() got an unexpected keyword argument 'slots'

    /path/to/mambaforge/mambaforge/envs/humble-msa/lib/python3.9/site-packages/moveit_configs_utils/moveit_configs_builder.py Line 84:

    @dataclass(slots=True)
    class MoveItConfigs:
        """Class containing MoveIt related parameters."""
    
        # A pathlib Path to the moveit config package
        package_path: Optional[str] = None
        # A dictionary that has the contents of the URDF file.
        robot_description: Dict = field(default_factory=dict)
    ... etc.

    There's probably something more subtle that could happen for a patch beyond just disabling the slots ... will report back if that works. It looks like manually declaring __slots__ conflicts with setting the default values.

    5 replies
    Silvio Traversaro
    @traversaro:matrix.org
    [m]
    The use of dataclasses was added in ros-planning/moveit2#1308, jafar that contributed it if I am not wrong partecipated in a robostack meeting
    this seems also a related issue: ros-planning/moveit2#1449
    and it seems that they are working on this upstream: ros-planning/moveit2#1450