Windows Installation #487
Description
Hi,
Just raising something I've noticed with the Windows helm install command. When I run this on my machine, I receive an error regarding sh not being executable - as I'd expect from a Windows machine without Cygwin/WSL interactions. I was able to get around this by installing the plugin in this method then copying the Windows release from Github into \bin within the plugin directory.
It isn't possible to run helm plugin install ~\Downloads\helm-diff-windows-amd64.tgz since this is just the built parts of the plugin.
According to the debug information for the installation:
[debug] vcs_installer.go:162: updating https://github.com/databus23/helm-diff
[debug] vcs_installer.go:91: copying C:\Users\JimSpeir\AppData\Local\Temp\helm\plugins\https-github.com-databus23-helm-diff to C:\Users\JimSpeir\AppData\Roaming\helm\plugins\helm-diff
plugin_install.go:82: [debug] loading plugin from C:\Users\JimSpeir\AppData\Roaming\helm\plugins\helm-diff
plugin.go:60: [debug] running install hook: sh -c $HELM_PLUGIN_DIR/install-binary.sh
Error: exec: "sh": executable file not found in %PATH%
helm.go:84: [debug] exec: "sh": executable file not found in %PATH%
I think it should be relatively trivial to mimic the install-binary.sh script to work as a native PowerShell script. Would there be any issues with this approach or is there a reason the installation is done using sh? I think we could safely get around OS/architecture constraints using the platformCommand list within plugin.yaml.