This is the official command line client for Cloud Foundry.
You can follow our development progress on Pivotal Tracker.
Download and run the installer for your platform from the Downloads Section.
Once installed, you can log in and push an app.
$ cd [my-app-directory]
$ cf api api.[my-cloudfoundry].com
Setting api endpoint to https://api.[my-cloudfoundry].com...
OK
$ cf login
API endpoint: https://api.[my-cloudfoundry].com
Email> [my-email]
Password> [my-password]
Authenticating...
OK
$ cf push
#Further Reading and Getting Help
- You can find further documentation at the docs page for the CLI here.
- There is also help available in the CLI itself; type
cf help
for more information. - Each command also has help output available via
cf [command] --help
orcf [command] -h
. - For development guide on writing a cli plugin, see here.
- Finally, if you are still stuck or have any questions or issues, feel free to open a GitHub issue.
WARNING: Edge binaries are published with each new 'push' that passes though CI. These binaries are not intended for wider use; they're for developers to test new features and fixes as they are completed.
Stable Installers | Stable Binaries | Edge Binaries |
---|---|---|
Mac OS X 64 bit | Mac OS X 64 bit | Mac OS X 64 bit |
Windows 32 bit | Windows 32 bit | Windows 32 bit |
Windows 64 bit | Windows 64 bit | Windows 64 bit |
Redhat 32 bit | Linux 32 bit | Linux 32 bit |
Redhat 64 bit | Linux 64 bit | Linux 64 bit |
Debian 32 bit | ||
Debian 64 bit |
Experimental: Install CF for OSX through Homebrew via the pivotal's homebrew-tap:
$ brew tap pivotal/tap
$ brew install cloudfoundry-cli
Releases: Information about our releases can be found here
- .cfignore used in
cf push
must be in UTF8 encoding for CLI to interpret correctly.
- "bash: .cf: No such file or directory". Ensure that you're using the correct binary or installer for your architecture. See http://askubuntu.com/questions/133389/no-such-file-or-directory-but-the-file-exists
- the command you ran
- what occurred
- what you expected to occur
For bugs related to HTTP requests or strange behavior, please run the command with env var CF_TRACE=true
and provide
- the command you ran
- the trace output
- a high-level description of the bug
- the command you ran
- the stack trace generated (if any)
- any other relevant information
- Install Go
- Ensure your $GOPATH is set correctly
- Install godep
- Get the cli source code:
go get github.com/cloudfoundry/cli
- (Ignore any warnings about "no buildable Go source files")
- Run
godep restore
(note: this will modify the dependencies in your $GOPATH) - Fork the repository
- Add your fork as a remote:
cd $GOPATH/src/github.com/cloudfoundry/cli && git remote add your_name https://github.com/your_name/cli
To prepare your build environment, run go get github.com/jteeuwen/go-bindata/...
- Run
./bin/build
- The binary will be built into the
./out
directory.
Optionally, you can use bin/run
to compile and run the executable in one step.
- Install Mercurial
- Run
go get golang.org/x/tools/cmd/vet
- Write a Ginkgo test.
- Run
bin/test
and watch the test fail. - Make the test pass.
- Submit a pull request to the
master
branch.
* For development guide on writing a cli plugin, see here
Major new feature proposals are given as a publically viewable google document with commenting allowed and discussed on the vcap-dev mailing list.
Pull Requests should be made against the master
branch.
A command is a struct that implements this interface:
type Command interface {
Metadata() command_metadata.CommandMetadata
GetRequirements(requirementsFactory requirements.Factory, c *cli.Context) (reqs []requirements.Requirement, err error)
Run(c *cli.Context)
}
Metadata()
is just a description of the command name, usage and flags:
type CommandMetadata struct {
Name string
ShortName string
Usage string
Description string
Flags []cli.Flag
SkipFlagParsing bool
}
GetRequirements()
returns a list of requirements that need to be met before a command can be invoked.
Run()
is the method that your command implements to do whatever it's supposed to do. The context
object
provides flags and arguments.
When the command is run, it communicates with api using repositories (they are in cf/api
).
Dependencies are injected into each command, so tests can inject a fake. This means that dependencies are
typically declared as an interface type, and not a concrete type. (see cf/commands/factory.go
)
Some dependencies are managed by a repository locator in cf/api/repository_locator.go
.
Repositories communicate with the api endpoints through a Gateway (see cf/net
).
Models are data structures related to Cloud Foundry (see cf/models
). For example, some models are
apps, buildpacks, domains, etc.
Command dependencies are managed by the commands factory. The app uses the command factory (in cf/commands/factory.go
)
to instantiate them, this allows not sharing the knowledge of their dependencies with the app itself.
As for repositories, we use the repository locator to handle their dependencies. You can find it in cf/api/repository_locator.go
.
Create Space is a good example of a command. Its tests include checking arguments, requiring the user
to be logged in, and the actual behavior of the command itself. You can find it in cf/commands/space/create_space.go
.
All pull requests which include user-facing strings should include updated translation files. These files are generated/ maintained using i18n4go.
To add/ update translation strings run the command i18n4go -c fixup
. For each change or update, you will be presented with the choices new
or upd
. Type in the appropriate choice. If upd
is chosen, you will be asked to confirm which string is being updated using a numbered list.
Resources that include several commands have been broken out into their own sub-package using the Resource name. An example
of this convention is the Space resource and package (see cf/commands/space
)
In addition, command file and methods naming follows a CRUD like convention. For example, the Space resource includes commands such a CreateSpace, ListSpaces, DeleteSpace, etc.
Although not ideal, we use the name "Repository" for API related operations as opposed to "Service". Repository was chosen to avoid confusion with Service model objects (i.e. creating Services and Service Instances within Cloud Foundry).
By convention, Repository methods return a model object and an error. Models are used in both Commands and Repositories to model Cloud Foundry data. This convention provides a consistent method signature across repositories.