DockerImage could not find skbuild

[joshi-bharat]

While running the docker command, pip could not find skbuild.
Looks like skbuild is replaced with scikit-build.

Looks like it's coming from OpenCV.4.3.0 build during benchbot_api build.
Important, this is inherent in opencv-python install. Not the problem of this repo.

Just leaving here for reference.

[btalb]

Good catch @joshi-bharat; thanks for sharing it with us here.

I was coincidentally running an installation last night which also failed with exactly the same error. The error seems to have been introduced by a version of python-opencv released in the last day or so. Your suggestion was really helpful in nailing down the root of the issue:

• Python OpenCV now requires manylinux2014 to build the wheel for the package (Pip install is broken (missing requirement) opencv/opencv-python#372 (comment)), which has been available since v19.3 of pip
• Ubuntu's python3-pip package installs v9.x of pip.... which is over 2.5 years old at this point
• Upgrading pip under Ubuntu via pip3 install --upgrade pip causes pip to throw warnings
• It is impossible to get rid of the warnings without setting up virtual environments, or manually deleting system files (ImportError in system pip wrappers after an upgrade pypa/pip#5599)...

I have added fixes to the two places we install the API: creation of the submission Dockerfile, & installation on the host system. The fixes will likely throw the following warning when using pip:

WARNING: pip is being invoked by an old script wrapper. This will fail in a future version of pip. Please see https://github.com/pypa/pip/issues/5599 for advice on fixing the underlying issue. To avoid this problem you can invoke Python with '-m pip' instead of running pip directly.

It's just a warning, but still something I wish wasn't there. On your host system I'd recommend having a recent version of pip & some sort of virtual environment setup regardless which will also get around the warnings.

See commits 42f5ba8 & df078ba for the fixes, which are available in the v1.0.4 release. Let me know if you have any further issues.

(I'd be lying if I said I thought it was okay for something as ubiquitous as pip to require anything more than sudo apt install python3-pip to get running without issues.... but we won't go into that 😛 )