gcloud_topic_command-conventions (1)

NAME

gcloud topic command-conventions - gcloud command conventions supplementary help

DESCRIPTION

gcloud command design follows a common set of principles and conventions. This document describes them in detail.
Conventions are goals more than rules. Refer to individual command --help for any exceptions.

Command Hierarchy

gcloud commands are organized as a tree with gcloud at the root, command groups in the inner nodes, and commands at the leaf nodes. Each command group typically contains a set of CRUD commands (create, describe, list, update, delete) that operate on a resource for a single API. Group commands are executable, but only for displaying help.
All groups and commands have a --help flag that displays a man(1) style document on the standard output. The display is run through the default pager if the calling environment specifies one. Help documents are derived from the running executable, so they are always up to date, even when switching between multiple release installations.

Command Line

Every gcloud command line follows the same form:

gcloud GROUP GROUP ... COMMAND POSITIONAL ... FLAG ...

Flag and positional arguments can be intermixed but for consistency are usually displayed positionals first in order, followed by flags in any order.

Command Usage Notation

Command usage is a shorthand notation that contains the full command name, the positional arguments, and the flag arguments in group sorted order. Optional arguments are enclosed in [ ... ]. For example:

gcloud foo bar NAME [EXTRA] [--format=FORMAT]

is the usage for the gcloud foo bar command with a required NAME positional argument, an optional EXTRA positional argument, and an optional --format flag argument.
Mutually exclusive arguments are separated by |; at most one arg in the list of mutually exclusive args may be specified:

[ --foo | --bar ]

This means that either --foo or --bar may be specified, but not both.
Mutually exclusive args may also be required, meaning exactly one arg in the list must be specified. This is denoted by enclosing the args in ( ... ):

( --foo | --bar )

Modal argument groups are also supported. If any arg in the group is specified, then the modal arguments must also be specified. This is denoted by using : to separate the modal args on the left from the other args on the right:

[ --must-a --must-b : --maybe-c --maybe-d ]

This means that if --maybe-c and/or --maybe-d are specified then both --must-a and --must-b must be specified.

Positional Arguments

Positional arguments are ordered and must be specified in the order listed in the command usage and help document argument definition list.
File input arguments usually accept the special name - to mean read from the standard input. This can be used only once per command line.

Flag Arguments

Flag names are lower case with a -- prefix. Multi-word flags use - (dash) as a word separator. Single character flags are deprecated, rare and may not be documented at all.
Following UNIX convention, if a flag is repeated on the command line, then only the rightmost occurrence takes effect, no diagnostic is emitted. This makes it easy to set up command aliases and wrapper scripts that provide default flag values; values that can easily be overridden by specifying them on the alias or wrapper script command line.

Boolean Flags

Boolean flags have an implied value of false or true. The presence of --foo sets the flag to true. All Boolean flags have a --no- prefix variant. For example, --no-foo sets the Boolean --foo flag to false. Boolean flags are documented using the positive form. This keeps the style consistent across all commands, and also makes the meaning of the --no- variant clear. In the case a Boolean flag has a default value of true, the --no- variant will appear in the command usage and help text and like all other --no- flags, will set the value of the flag to false.

Valued Flags

Non-Boolean flags have an explicit value. The value can be specified using =:

--flag=value

or by placing the value as the next arg after the flag:

--flag value

The = form must be used if value starts with -.
The second form requires extra context to determine if --flag is Boolean and value is a positional, or if --flag is valued and value is its value. Because of the visual ambiguity, usage notation and most command examples use the first form to make intentions clear. The = form also has a diagnostic bonus: it is an error to specify a value for a Boolean flag.

Complex Flag Values

Complex flag values that contain command interpreter special characters may be difficult to specify on the command line. The --flags-file=YAML-FILE flag solves this problem by allowing command line flags to be specified in a YAML/JSON file. String, numeric, list and dict flag values are specified using YAML/JSON notation and quoting rules. See $ gcloud topic flags-file for more information.

Output

The standard output is for explicit information requested by the command. Depending on the context, there may be guarantees on the output format to support deterministic parsing. For example, the yaml, json and csv output --formats guarantee that successful command completion results in standard output data that can be parsed using the respective format.
The standard error is reserved for diagnostics. In general, the format of standard error data may change from release to release. Users should not script against specific content, or even the existence of output to the standard error at all. The only reliable error indicator is the exit status described below.
Most standard error messaging is also logged to a file that can be accessed by $ gcloud info --show-log.
No gcloud command should crash with an uncaught exception. However, if gcloud does crash the stack trace is intercepted and written to the log file, and a crash diagnostic is written to the standard error.

Exit Status

Exit status 0 indicates success. For async commands it indicates that the operation started successfully but may not have completed yet.
Any other exit status indicates an error. Command-specific diagnostics should explain the nature of the error and how to correct it.