Shorten the number of steps to run pulumi up

[hausdorff]

As I was testing StatefulSet I realized that my default when I want to "just try" something is to:

1. Copy a bit of YAML into a file.
2. kubectl apply -f

rather than to (e.g.):

1. Run pulumi new.
2. Copy a bit of YAML into a file.
3. Write k8s.yaml.ConfigFile for that file.
4. pulumi up

It would be great to make it as easy to use Pulumi as it is to use kubectl. Some ideas off the top of the head:

• Allow pulumi new to take a file/URL to Kubernetes manifests which creates an app that points at the relevant YAML.
• Create a pulumi convert which will generate a "real" Pulumi program based on the YAML.

cc @joeduffy @lukehoban @lblackstone

[brandonkal]

For the simple "try kubernetes things out" case, one idea is that YAML could have its own simple language provider that would not require installing node modules and other project boilerplate. Instead, the user adds a #@pulumi pragma to the top of the yaml file and runs pulumi up.

To take it one step further, Pulumi.yaml project config could be defined as a custom resource that does not get applied to the cluster, like the Sources resource in kbld to reduce files to a single portable multi-doc file.

[rshade]

Would this work for you problem? https://github.com/pulumi/pulumi-yaml/blob/main/examples/kubernetes/Pulumi.yaml