Dataflow could not be validated for examples

[jmskov]

Describe the bug
I have installed dora according to the instructions and Getting Started section in the README. I can build the example python-dataflow fine. I get the error [ERROR] Dataflow could not be validated. when running the example.

To Reproduce
Steps to reproduce the behavior:

0. Install dora-cli using cargo, activate venv in examples/python-dataflow
1. Run dora up
2. Run dora build dataflow.yaml
3. Run dora start dataflow.yaml

This pattern continues for other python examples.

Expected behavior
The webcam example to run.

Environments (please complete the following information):

• System info: MacOS 15.1, M2Pro.
• Dora version: dora-cli 0.3.6,
  Name: dora-rs
  Version: 0.3.6
  Summary: dora goal is to be a low latency, composable, and distributed data flow.
  Home-page:
  Author:
  Author-email:
  License: Apache-2.0
  Location: /Users/john/Projects/rust/dora/dora/examples/python-dataflow/meow/lib/python3.12/site-packages
  Requires: pyarrow
  Required-by: dora-yolo, opencv-plot, opencv-video-capture

Additional context
Using Python 3.12.7 and pip via homebrew. I can confirm import cv2; cap = cv2.VideoCapture(0) can capture my webcam from the interpreter in the venv.

[haixuanTao]

Ok, interesting, I wonder if it's because, the environment path management for python in macOS is different.

Going to double check this and probably change the example.

Thanks for reporting this issue.

[haixuanTao]

Do you mind trying out the v0.3.7rc0 version?

You can install the new version with:

curl --proto '=https' --tlsv1.2 -sSf https://raw.githubusercontent.com/dora-rs/dora/main/install.sh | bash -s -- --tag v0.3.7rc0

source ~/.bashrc

Or download it manually at: https://github.com/dora-rs/dora/releases/tag/v0.3.7rc0

[jmskov]

Same result, thanks for the suggestion. Also not sure if it is an oversight, but the crate version was not changed in the release candidate. I will try some other ideas and update if I make progress.

[haixuanTao]

Ok sorry.

Yeah, I wanted to do a quick release candidate without having to change all package version conflict. Maybe v0.3.6-post naming would have made more sense.

Those path dependency are very hard to predict.

I want to add a test suite within dora to simplify testing.

Basically, you can try to do:

pip install opencv-video-capture opencv-plot
# And try individually:
opencv-video-capture

# It it returns a path error it means that the installation didn't work
# If it returns an error with `DORA_NODE_CONFIG` it means that it should work, but the path couldn't be validated.

[jmskov]

No worries, just wanted to check 😄 I believe my error is the latter, here is my trace when running opencv-video-capture:

File "/Users/john/Projects/rust/dora/dora/examples/python-dataflow/bin/opencv-video-capture", line 8, in <module>
    sys.exit(main())
             ^^^^^^
  File "/Users/john/Projects/rust/dora/dora/examples/python-dataflow/lib/python3.12/site-packages/opencv_video_capture/main.py", line 74, in main
    node = Node(args.name)
           ^^^^^^^^^^^^^^^
RuntimeError: Could not setup node from node id. Make sure to have a running dataflow with this dynamic node

Caused by:
    failed to get node config from daemon: no node with ID `{node_id}`

Location:
    apis/rust/node/src/node/mod.rs:106:17

[haixuanTao]

Ok, so I think the path resolver from the cli did not work, which is somehow surprising. As it should work ... in theory.

Would you mind doing the following:

dora destroy 
dora up
dora start dataflow.yaml

[jmskov]

That fixed it. I tried a few permuations to replicate the issue (e.g. dora up in new terminal, outside of the env), but I have not been able to.