Allow Root commands when there are subcommands

[F21]

If I have a CLI with sub commands, it is not possible to call the root comand, even if Run() is defined at the root.

For example:

type CLI struct {
	SubCmd1 SubCmd1 `cmd:""`
	SubCmd2 SubCmd2  `cmd:""`
}

func (cli *CLI) Run() error {
	fmt.Println("This is the root command")
	return nil
}

type SubCmd1 struct {}

func (c *SubCmd1) Run() error {
	fmt.Println("This is sub command 1")
	return nil
}

type SubCmd2 struct {}

func (c *SubCmd3) Run() error {
	fmt.Println("This is sub command 3")
	return nil
}

func main() {
	var cli CLI
	ctx := kong.Parse(&cli)
	err := ctx.Run()
	ctx.FatalIfErrorf(err)
}

Upon compiling, it is not possible to call ./myapp as it will return the following error: myapp: error: expected one of "sub-cmd1", "sub-cmd2". I would love to have the ability to call Run() on the root: ./myapp as well as the sub commands: ./myapp sub-cmd1.

[christhomas]

I'm also running into this and I can't figure out what I'm supposed to do, it seems that once you declare any command, you are obligated to use them and you can't do something like --version or --help on the application root itself anymore.

[alecthomas]

@christhomas I'm not sure what you're doing exactly, but --help works fine in that scenario

[sewnie]

The current behavior is better. You don't want to run other commands on accident. Nested subcommands help with the command's arguments, and how they're related. If you need to do other things, I think it is better you use the available command hooks.

[alecthomas]

This wouldn't necessarily be a backwards incompatible change. It could be a case of "if there's a Run() method then that inner node can be a terminating command".