databricks / cli Goto Github PK

We removed GPG signing because hashicorp deprecated their GH action and it was already disabled in .goreleaser.yaml.

removed from release.yaml:

      - name: Import GPG key
        id: import_gpg
        uses: crazy-max/[email protected]
        with:
          gpg_private_key: ${{ secrets.GPG_PRIVATE_KEY }}
          passphrase: ${{ secrets.GPG_PASSPHRASE }}

removed from .goreleaser.yaml:

signs:
  - artifacts: checksum
    args:
      - "--local-user"
      - "{{ .Env.GPG_FINGERPRINT }}"
      - "--output"
      - "${signature}"
      - "--detach-sign"
      - "${artifact}"

Lets start adding unit tests for the bricks cli commands. We can start with bricks fs ls ... since that should be the easiest

and run unit tests on windows

Right now it's hardcoded in the vscode extension to only validate databricks.yml and bundle.yml files with the json schema

The SDK has mocking support and can be leveraged in bricks tests.

That's compatible with the equivalent databricks configure command in the existing CLI

https://docs.databricks.com/dev-tools/cli/index.html#set-up-authentication

Please fix it :)

see: https://github.com/databricks/bricks/runs/8209395948?check_suite_focus=true

see: https://databricks.slack.com/archives/C03AV2J5FSR/p1663848308249659

Remove CreateDbfsFile and ReadDbfsFile once equivalent helper functions are implemented in go SDK. Tracked in go SDK in issue: databricks/databricks-sdk-go#57

Currently when running ./bricks bundle validate I am unable to determine the auth method decided by the bricks cli and the workspace url.

If validate is run i would like to debug environment targets (URL and auth mechanism) being selected in the logs incase there are any permission issues.

Potentially somewhere in the workspace block:

This will also require support for pagination on the bricks client side

This will allow us to avoid an unnecessary error: see: #48 (comment)

low priority as of 13 Sep 2022

Remote repo projects will slowly drift from local files because of two reasons:

1. Deleting dirs locally does not delete files in the workspace repo
2. Files deleted when bricks sync was not running will not be deleted.

This can pollute the remote repo

When this flag is provided, we do not persist the in-memory snapshot to local disk

Two fields in resources.Pipeline have the json tag id. We should also remove one of them

Running gotestsum shows failures in multiple packages.

https://github.com/fsnotify/fsnotify

see: #48 (comment)

see dbx implementation for how to do it

Remove GetAllRepos once equivalent helper function is implemented in go SDK

Tracked in go SDK repo in issue: databricks/databricks-sdk-go#58

Apparently, it exists

Syntax runs, validation errors, expectation fails etc

Replace use of panic(err) with the RunE callback

#19 (comment) is one instance of such a use

Right now if you change the destination workspace, your terraforms state gets nuked. We should have a warning/error if a deploy would be destructive to remote state.

Does not apply when the state is remote. However we can potentially support a local only terraform state location handling

Then have the io.Reader and io.Writer inside the context? Can be called then like cmdio.Log, cmdio.Prompt...

Add the following arguments to bricks configure to get better parity with the Databricks CLI.

  -t, --token
  --host TEXT                     Host to connect to.
  --profile TEXT                  CLI connection profile to use. The default
                                  profile is "DEFAULT".

Either by doing the first upload as a zip or by concurrent upload requests

Even though tests pass, the GitHub action fails