Without this the build will freeze with prompt:
Proceed ([y]/n)?
I'm using rootless podman in vscode instead of docker but I think it should not affect this.
..or does conda somehow detect Docker but not Podman? Anyway, this should not break anything.
Btw, I also had to uncomment the line: "remoteUser": "root" in devcontainer.json to finish the post installation properly but I guess there might be other workarounds - and perhaps you don't want to run as root if your container has root privileges.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/121128
Approved by: https://github.com/drisspg
Building with CUDA in dev container leads to error: `cannot find -lcudart_static`. This is because the libraries are under a custom CUDA_HOME, and `ld` cannot find it.
Updating the `LDFLAGS` environment variable works.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/108766
Approved by: https://github.com/drisspg
# Summary
<!--
copilot:summary
-->
### <samp>🤖 Generated by Copilot at 293ded1</samp>
This pull request adds support for using Visual Studio Code Remote - Containers extension with the pytorch project. It adds a `.devcontainer` folder with a `devcontainer.json` file, a `Dockerfile`, and a `noop.txt` file that configure and create a dev container with Anaconda and Python 3.
<!--
copilot:poem
-->
### <samp>🤖 Generated by Copilot at d6b9cd7</samp>
> _`devcontainer.json`_
> _Configures PyTorch containers_
> _For CPU or GPU_
## Related to:
https://github.com/pytorch/pytorch/issues/92838
Pull Request resolved: https://github.com/pytorch/pytorch/pull/98252
Approved by: https://github.com/ZainRizvi
