Recover from panic gracefully

[shreyas-goenka]

Changes

This PR adds a recovery function for panics. This indicates to all users running into a panic that it's a bug and they should report it to Databricks.

Tests

Manually. I added panic(fmt.Errorf("some error message")) to the ConfigureBundleWithVariables function which is the fourth entry in the stack trace.

Error message:

.venv➜  cli git:(panic-r) ✗ ./cli bundle deploy
Error: The Databricks CLI unexpectedly panicked. This should not happen.
Please report this issue to Databricks in the form of a GitHub issue at:
https://github.com/databricks/cli

CLI Version: 0.0.0-dev+2d0963661136

Panic Payload: some error message

Stack Trace:
goroutine 1 [running]:
runtime/debug.Stack()
        /Users/shreyas.goenka/go/pkg/mod/golang.org/[email protected]/src/runtime/debug/stack.go:26 +0x64
github.com/databricks/cli/cmd/root.Execute.func1()
        /Users/shreyas.goenka/cli2/cli/cmd/root/root.go:110 +0xa4
panic({0x103408c40?, 0x14000116260?})
        /Users/shreyas.goenka/go/pkg/mod/golang.org/[email protected]/src/runtime/panic.go:785 +0x124
github.com/databricks/cli/cmd/bundle/utils.ConfigureBundleWithVariables(...)
        /Users/shreyas.goenka/cli2/cli/cmd/bundle/utils/utils.go:21
github.com/databricks/cli/cmd/bundle.newDeployCommand.func1(0x14000164c08?, {0x1040ff1c0?, 0x4?, 0x1030601e7?})
        /Users/shreyas.goenka/cli2/cli/cmd/bundle/deploy.go:44 +0x4c
github.com/spf13/cobra.(*Command).execute(0x14000164c08, {0x1040ff1c0, 0x0, 0x0})
        /Users/shreyas.goenka/cli2/cli/vendor/github.com/spf13/cobra/command.go:985 +0x834
github.com/spf13/cobra.(*Command).ExecuteC(0x14000414608)
        /Users/shreyas.goenka/cli2/cli/vendor/github.com/spf13/cobra/command.go:1117 +0x344
github.com/spf13/cobra.(*Command).ExecuteContextC(...)
        /Users/shreyas.goenka/cli2/cli/vendor/github.com/spf13/cobra/command.go:1050
github.com/databricks/cli/cmd/root.Execute({0x1036e3e00?, 0x1040ff1c0?}, 0x103fd1d68?)
        /Users/shreyas.goenka/cli2/cli/cmd/root/root.go:125 +0x88
main.main()
        /Users/shreyas.goenka/cli2/cli/main.go:13 +0x44

[denik]

Could we add an acceptance test by either forcing panic somehow of add special flag to inject it?

[shreyas-goenka]

Could we add an acceptance test by either forcing panic somehow or adding a special flag to inject it?

We could, but I think it's fine to skip this one. There is no great way to do this since we would have to add a new dummy command or flag just for this.

The goal is to have a QOL improvement for customers and make it clear to them that this is a bug on our side. Even if it fails we are not losing out on anything crucial.

[pietern]

Could you include an acceptance test for the output?

I'd like to see the diff between the current panic output and the new panic output.

[denik]

Panic is a golang-specific term that users are not required to know, perhaps "The Databricks CLI had fatal error"?

[shreyas-goenka]

Either way works. The next line does direct the user to open an issue on Databricks so they don't need to know what panicking means.

[denik]

(see inline comments)