Make sure Cobra is configured with the streams we use, and use
Cobra's utilities to print the validation messsage.
While updating, also add a short comment outlining why we're using
STDERR, not STDOUT for this message.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This workaround was added in 3bb40af036 to
work around a limitation of Cobra, which didn't support completion if there
are no subcommands, because [Command.InitDefaultCompletionCmd] disabled
completion, as implicitly adding the hidden `completion` and `__complete`
subcommands would break binaries without subcommands that accepted positional
args.
This problem was fixed in Cobra v1.9.0 through [cobra@24ada7f], which now
dynamically adds the `completion` subcommand when invoked.
This patch removes the hidden `__dummy_command`, which was used to work around
the logic in Cobra < v1.9.0, and to make it enable completion regardless.
Completion should still work as usual:
dockerd completion --help
Usage: dockerd completion [flags]
...
source <(dockerd completion bash)
dockerd --default-<tab>
--default-address-pool (Default address pools for node specific local networks)
--default-cgroupns-mode (Default mode for containers cgroup namespace ("host" | "private"))
--default-gateway (Default gateway IPv4 address for the default bridge network)
--default-gateway-v6 (Default gateway IPv6 address for the default bridge network)
--default-ipc-mode (Default mode for containers ipc ("shareable" | "private"))
--default-network-opt (Default network options)
--default-runtime (Default OCI runtime for containers)
--default-shm-size (Default shm size for containers)
--default-ulimit (Default ulimits for containers)
[Command.InitDefaultCompletionCmd]: https://github.com/spf13/cobra/blob/v1.8.1/completions.go#L685-L698
[cobra@24ada7f]: https://github.com/spf13/cobra/commit/24ada7fe71e3a3a8741dd52e0a7fc3b97450535a
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This package is used internally by the daemon, and was only used out
side of the daemon by pkg/plugins (for which we still need to look
where it should be kept).
Making it internal because it's trivial to implement if needed by
anyone. The only reason it's a package is to keep it central, and
to make it easier to discover where we have rootlesskit-specific
codepaths in our codebase.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Sebastiaan van Stijn