Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
    Yong He
    @heyong4725
    @gabrik , we are trying to compile zenoh flow on Ubuntu 18.04 server, we can build zenoh flow runtime without issue. And also we could build fizz example. However, we cannot compile open CV examples (following your readme instruction). Wonder if you have similar problem during your experiment process:
    image.png
    image.png
    I guess this is opencv installation problem, but we did not encounter any opencv installation issues.
    You may give us some hints. we are trying to reproduce your demo today
    J-Loudet
    @J-Loudet
    Hi @heyong4725, @gabrik is on holidays so he won’t be available for a while. Unfortunately, at the moment I only have at my disposal a Mac with the M1 chip (which proves to be an immense struggle in order to compile / link OpenCV) so I am unable to help. Tomorrow I will have access to a linux laptop and I will try to reproduce your error. Can you tell me which version of OpenCV you are running on the server?
    Gabriele Baldoni
    @gabrik

    Hi @heyong4725 I guess it has to do with the version of opencv, the rust bindings are made to work with opencv 4.x, can you check the version that comes with Ubuntu 18.04? apt list --installed | grep libopencv-dev

    This is the version we have in Ubuntu 20.04:

    $ apt list --installed | grep libopencv-dev
    
    WARNING: apt does not have a stable CLI interface. Use with caution in scripts.
    
    libopencv-dev/focal,now 4.2.0+dfsg-5 arm64 [installed]
    Bob Ding
    @bobd988
    Hi @gabrik , when I run any of example like cargo build --example manual-source --example example-fizz --example example-buzz --example generic-sink , then I got error message ./target/release/runtime --graph-file zenoh-flow-examples/graphs/fizz_buzz_pipeline.yaml --runtime foo
    thread 'main' panicked at 'called Result::unwrap() on an Err value: LoadingError("Error when loading the library: ./target/debug/examples/libexample_fizz.so: undefined symbol: _ZN2cv3sfm11reconstructESt6vectorINSt7__cxx1112basic_stringIcSt11char_traitsIcESaIcEEESaIS7_EERKNS_12_OutputArrayESC_RKNS_17_InputOutputArrayESC_b")', zenoh-flow/src/bin/runtime.rs:70:39
    note: run with RUST_BACKTRACE=1 environment variable to display a backtrace
    The so file exists in right path defined in yaml.
    Gabriele Baldoni
    @gabrik
    Hi @bobd988 it looks like an ABI error, can you try to rebuild the runtime and then the examples?
    Bob Ding
    @bobd988

    @gabrik I tried again to clean and build project and then each example. Unfortunately I got same error. Second option I tried is to get latest code and build. There is error message Compiling opencv v0.53.1
    Compiling zenoh-flow-derive v0.0.1 (~/zenoh-flow-latest/zenoh-flow-derive)
    Compiling zenoh v0.5.0-dev (https://github.com/eclipse-zenoh/zenoh.git?branch=master#f97c9bc2)
    error[E0433]: failed to resolve: use of undeclared type SharedMemoryAuthenticator
    --> ~/.cargo/git/checkouts/zenoh-cc237f2570fab813/f97c9bc/zenoh/src/net/transport/unicast/manager.rs:193:25
    |
    193 | .insert(SharedMemoryAuthenticator::new().into());
    | ^^^^^^^^^^^^^^^^^^^^^^^^^ use of undeclared type SharedMemoryAuthenticator

    error: aborting due to previous error

    For more information about this error, try rustc --explain E0433.
    error: could not compile zenoh

    Enea Scioni
    @haianos
    @bobd988 I had the same was fixed in eclipse-zenoh/zenoh-flow#7
    Gabriele Baldoni
    @gabrik
    Hi @bobd988, can you try with:
    git pull
    cargo clean
    cargo update
    cargo build --all-targets
    Bob Ding
    @bobd988
    with latest pull and clean these above errors gone. I can run opencv with CUDA single runtime example successfully. When I run multi-runtime, for the first command, it gives error : (base) ./target/release/runtime --graph-file zenoh-flow-examples/graphs/dnn-object-detection-multi-runtime.yaml --runtime foo
    thread 'main' panicked at 'called Result::unwrap() on an Err value: ParsingError("while parsing a block collection, did not find expected '-' indicator at line 16 column 3")', zenoh-flow/src/bin/runtime.rs:46:82
    note: run with RUST_BACKTRACE=1 environment variable to display a backtrace
    Aborted (core dumped)
    Gabriele Baldoni
    @gabrik
    Hi @bobd988 it should be a syntax error in the yaml file.
    I have tested with the one that comes in the repository and it is working, I’m copying it here for your convienience:
    flow: DNNPipeline # it should have a id, from this we will have an Instance of the graph, UUID => flow
    operators:
      - id : ObjectDetection # this should be unique in the graph
        uri: file://./target/release/examples/libobject_detection_dnn.so
        inputs:
          - id: Frame
            type: image
        outputs:
          - id: Frame
            type: image
        configuration:
          neural-network: /home/gabri/Workspace/experiments/yolov3-face.cfg
          network-weights: /home/gabri/Workspace/experiments/yolov3-wider_16000.weights
          network-classes: /home/gabri/Workspace/experiments/face_classes.txt
    sources: # sources can have only ONE output
      - id : Camera0
        uri: file://./target/release/examples/libcamera_source.so
        output:
          id: Frame
          type: image
        configuration:
          camera: /dev/video0
          resolution: 800x600
          fps: 24
    sinks: # sinks can have only ONE input
      - id : Window
        uri: file://./target/release/examples/libvideo_sink.so
        input:
          id: Frame
          type: image
    
    links:
    - from: Camera0.Frame
      to: ObjectDetection.Frame
    - from: ObjectDetection.Frame
      to: Window.Frame
    
    mapping:
      - id: Window
        runtime: foo
      - id: Camera0
        runtime: bar
      - id: ObjectDetection
        runtime: cuda
    bob ding
    @bobd988_gitlab
    thanks. this works

    I am trying to build and run in Jetson Xavier with Ubuntu18.04. Rust installed is 1.51.0. when I build runtime it gave following error . I have installed LLVM and clang-7. Is there any suggestions ? Compiling quote v1.0.9
    Compiling unicode-normalization v0.1.19
    error: failed to run custom build command for clang-sys v1.2.2

    Caused by:
    process didn't exit successfully: /home/nvidia/Public/zenoh-flow/target/release/build/clang-sys-a9cebc8f04ee71c6/build-script-build (exit code: 101)
    --- stderr
    thread 'main' panicked at 'called Result::unwrap() on an Err value: "couldn\'t find any valid shared libraries matching: [\'libclang.so\', \'libclang-*.so\'], set the LIBCLANG_PATH environment variable to a path where one of these files can be found (invalid: [])"', /home/nvidia/.cargo/registry/src/github.com-1ecc6299db9ec823/clang-sys-1.2.2/build/dynamic.rs:211:45
    note: run with RUST_BACKTRACE=1 environment variable to display a backtrace
    warning: build failed, waiting for other jobs to finish...
    error: build failed

    bob ding
    @bobd988_gitlab
    in Xavier I got it fixed by uninstall clang related packages and then re-install with sudo apg-get install clang libclang-dev.
    Gabriele Baldoni
    @gabrik
    Hi @bobd988_gitlab , great that you solved, we do not have any Jetson powered SBC to test it, so we were trying to virtualize one.
    Thanks for sharing the solution!
    Gabriele Baldoni
    @gabrik
    Hi @heyong4725, we have just merged the cargo zenoh-flow command that facilitate users in writing their own components.
    Here you can find more information on how it works: https://github.com/eclipse-zenoh/zenoh-flow#creating-your-components
    Yong He
    @heyong4725
    @gabrik Thanks, I will give it a try next week
    Yong He
    @heyong4725
    @gabrik I followed https://github.com/eclipse-zenoh/zenoh-flow#creating-your-components instruction to generate the operate template, it seems there are some issues with latest check-ins.
    image.png
    It seems the template code has some conflict with latest check in from Olivier
    Gabriele Baldoni
    @gabrik
    @heyong4725 indeed, thanks for spotting this, we’ll fix right now
    Gabriele Baldoni
    @gabrik
    @heyong4725 it is now fixed by: eclipse-zenoh/zenoh-flow#22
    please let us know if you experience any other issue
    Gabriele Baldoni
    @gabrik
    We have also created a repository with the C++ Opencv example to show how complex operators can use the C++ api:
    https://github.com/atolab/zenoh-flow-cpp-opencv-example
    Yong He
    @heyong4725
    @gabrik thanks. @bobd988 can you try this C++ opencv example on your GPU laptop?
    Bob Ding
    @bobd988
    I built the so file from zenoh-flow-cpp-opencv-example. Then replaced the opencv dnn example libobject_detection_dnn.so , after ran I got below : ./target/release/runtime --graph-file zenoh-flow-examples/graphs/dnn-object-detection.yaml --runtime foo
    [ WARN:0] global /home/demo/Public/opencvbuild/opencv/modules/videoio/src/cap_gstreamer.cpp (1825) handleMessage OpenCV | GStreamer warning: Embedded video playback halted; module source reported: Could not read from resource.
    [ WARN:0] global /home/demo/Public/opencvbuild/opencv/modules/videoio/src/cap_gstreamer.cpp (914) open OpenCV | GStreamer warning: unable to start pipeline
    [ WARN:0] global /home/demo/Public/opencvbuild/opencv/modules/videoio/src/cap_gstreamer.cpp (501) isPipelinePlaying OpenCV | GStreamer warning: GStreamer: pipeline have not been created
    memory allocation of 4633403989658985 bytes failed
    Aborted (core dumped)
    1 reply
    Yong He
    @heyong4725
    @gabrik @J-Loudet Hi Gabriele, Julien, for our next Monday's sync-up meeting, I have invited AutoCore developers join our session to deep dive C++ APIs and their feedback and concerns from ROS2 developer's prospective. Please prepare some materials. Thanks
    This is ROS2 msg interfaces
    J-Loudet
    @J-Loudet
    @heyong4725 Perfect! We will indeed prepare some materials.
    Bob Ding
    @bobd988
    I get latest from repo. rebuilt opencv 4.5.2 with CUDA=ON , and verified the example object detection with dnn runs ok. Then I changed the yaml file to have the object detection so points to the libcxx_operator.so location. However I got error : ./target/release/runtime --graph-file zenoh-flow-examples/graphs/dnn-object-detection.yaml --runtime foo
    thread 'main' panicked at 'called Result::unwrap() on an Err value: VersionMismatch', zenoh-flow/src/bin/runtime.rs:70:39 . Then I checked the code is at this line in runtime.rs
    dataflow_graph.load(&opt.runtime).unwrap();
    Gabriele Baldoni
    @gabrik
    Hi @bobd988, can be solved by rebuilding the C++ example, the version mismatch error comes because it has found a different version of either rust or the rust core lib in the .so file/
    expploitt
    @expploitt
    Hello! Is it possible to migrate an specific operator from one node to another?
    Gabriele Baldoni
    @gabrik
    Hi @expploitt, not yet, the design supports it, but the functionality is not yet implemented.
    We are still focusing on the building blocks, more advance functionalities will come later on.
    Bob Ding
    @bobd988
    I got same error. Here is what I did 1. update to latest rust 1.55. 2. cargo install cxxbridge-cmd 3. removed c++ example dir and pull from repo, make libcxx_operator.so . 4. git pull and rebuild zenoh-flow runtime. 5. verify dnn works fine with object detection for single and multple runtime before any changes. 6. change yaml so to point to the libcxx_operator.so. Ubuntu 20.04 with CUDA 11.2 and driver 460.80, with only one version of opencv 4.5.2 installed in OS.
    Bob Ding
    @bobd988
    the c++ example works. I got it solved by removed all sudo apt-get purge 'opencv'
    sudo find / -name "opencv" -exec rm -rf {} \ , then rebuild all from source again with cuda 11.4 cudnn 8.2.2. Thanks
    Jiarre
    @Jiarre
    Hi, i'm new to zenoh.-flow and i'm getting errors trying to compile it.
    Is there something i'm missing from installation? Thanks in advance
    image.png
    2 replies
    Bob Ding
    @bobd988
    use this command to install dependencies: sudo apt-get install clang libclang-dev
    Gabriele Baldoni
    @gabrik
    Hi @bobd988 glad you solved it, I guess opencv sometimes causes troubles