Improvements
- `pulumi login` now supports `s3://`, `azblob://` and `gs://` paths (on top of `file://`) for
storing stack information. These are passed the location of a desired bucket for each respective
cloud provider (i.e. `pulumi login s3://mybucket`). Pulumi artifacts (like the
`xxx.checkpoint.json` file) will then be stored in that bucket. Credentials for accessing the
bucket operate in the normal manner for each cloud provider. i.e. for AWS this can come from the
environment, or your `.aws/credentials` file, etc.
- The pulumi version update check can be skipped by setting the environment variable
`PULUMI_SKIP_UPDATE_CHECK` to `1` or `true`.
- Fix an issue where the stack would not be selected when an existing stack is specified when running
`pulumi new <template> -s <existing-stack>`.
- Add a `--json` flag (`-j` for short) to the `preview` command. This allows basic serialization of a plan,
including the anticipated set of deployment steps, list of diagnostics messages, and summary information.
Each step includes deeply serialized information about the resource state and step metadata itself. This
is part of ongoing work tracked in [pulumi/pulumi2390](https://github.com/pulumi/pulumi/issues/2390).