This message resulted in code-lines that were too long; move it to a
const together with the other hint. While at it, also suppress unhandled
error, and touch-up the code-comment.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Deprecation comments must have an empty line before them, otherwise tools
and linters may not recognise them. While fixing this, also updated the
reference to PromptUserForCredentials to be a docs-link to make it clickable.
Updates 6e4818e7d6.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
If restoring the terminal state fails, "echo" no longer works, which means
that anything the user types is no longer shown. The login itself may already
have succeeded, so we should not fail the command, but it's good to inform
the user that this happened, which may give them a clue why things no longer
work as they expect them to work.
With this patch:
docker login -u yourname
Password:
Error: failed to restore terminal state to echo input: something bad happened
Login Succeeded
We should consider printing instructions how to restore this manually (other
than restarting the shell). e.g., 'run stty echo' when in a Linux or macOS shell,
but PowerShell and CMD.exe may need different instructions.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
we don't support empty passwords; when prompting the user for a password,
we already trim the result, but we didn't do the same for a password that's
passed through stdin or through the `-p` / `--password` flag.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
- move trimming defaultUsername inside the if-branch, as it's the only
location where the result of the trimmed username is use.
- do the reverse for trimming argUser, because the result of trimming
argUser is used outside of the if-branch (not just for the condition).
putting it inside the condition makes it easy to assume the result is
only used locally.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
move the "post" check for username being empty inside the branch
that's handling the username, as it's the only branch where username
is mutated after checking if it's empty.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
remove isDefaultRegistry and inline it where it's used; the code-comment
already outlines what we're looking for, so the intermediate var didn't
add much currently.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This function has multiple conditional branches, which makes it harder
to see at a glance whether authConfig may be partially populated. This
patch instead returns a fresh instance for error returns to prevent any
confusion.
It also removes the named output variables, as they're now no longer used,
and the returned types should already be descriptive enough to understand
what's returned.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
The Errors type is no longer used by the CLI itself, and this custom
"multi-error" implementation had both limitations (empty list not being
`nil`), as well as formatting not being great. All of this making it not
something to recommend, and better handled with Go's stdlib.
As far as I could find, there's no external consumers of this, but let's
deprecate first, and remove in the next release.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This command was using a custom "multi-error" implementation, but it
had some limitations, and the formatting wasn't great.
This patch replaces it with Go's errors.Join.
Before:
docker plugin remove one two three
Error response from daemon: plugin "one" not found, Error response from daemon: plugin "two" not found, Error response from daemon: plugin "three" not found
After:
docker plugin remove one two three
Error response from daemon: plugin "one" not found
Error response from daemon: plugin "two" not found
Error response from daemon: plugin "three" not found
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Prevent some tests from failing when running from a pre-compiled
testbinary, and discard output to make the output less noisy.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This commit implements a validation
method for the port mappings.
Also, it removes the ports validation
method from the expose property
since they do not accept the
same type of values.
Signed-off-by: Stavros Panakakis <stavrospanakakis@gmail.com>
When running tests from my IDE, it compiles the tests before running,
then executes the compiled binary to run the tests. Cobra doesn't like that,
because in that situation os.Args is taken as argument for the command that's
executed. The command that's tested now sees the `test-` flags as arguments
(`-test.v -test.run ..`), which causes various tests to fail ("Command XYZ
does not accept arguments").
# compile the tests:
go test -c -o foo.test
# execute the test:
./foo.test -test.v -test.run TestFoo
=== RUN TestFoo
Error: "foo" accepts no arguments.
Set arguments to an empty slice to make sure it doesn't inherit arguments
from the test-binary.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This test was added in [moby@b2551c6] as part of a larger PR that implemented
unit tests in various packages. In this specific test, it looks like the
`imageSaveFunc` that's defined in the test-table was forgotten to be wired
up, causing all tests to effectively be skipped.
This patch wires up the function so that it's used in the test.
[moby@b2551c6]: b2551c619d
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
With this patch, completion is provided for `--platform` flags:
docker pull --platform<TAB>
linux linux/amd64 linux/arm/v5 linux/arm/v7 linux/arm64/v8 linux/riscv64 wasip1 windows
linux/386 linux/arm linux/arm/v6 linux/arm64 linux/ppc64le linux/s390x wasip1/wasm windows/amd64
Note that `docker buildx build` (with BuildKit) does not yet provide completion;
it's provided through buildx, and uses a different format (accepting multiple
comma-separated platforms). Interestingly, tab-completion for `docker build`
currently uses completion for non-buildkit, and has some other issues that may
have to be looked into.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
With this patch, completion is provided for `--platform` flags:
docker run --platform<TAB>
linux linux/amd64 linux/arm/v5 linux/arm/v7 linux/arm64/v8 linux/riscv64 wasip1 windows
linux/386 linux/arm linux/arm/v6 linux/arm64 linux/ppc64le linux/s390x wasip1/wasm windows/amd64
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Add a utility for completing platform strings.
Platforms offers completion for platform-strings. It provides a non-exhaustive
list of platforms to be used for completion. Platform-strings are based on
[runtime.GOOS] and [runtime.GOARCH], but with (optional) variants added. A
list of recognised os/arch combinations from the Go runtime can be obtained
through "go tool dist list".
Some noteworthy exclusions from this list:
- arm64 images ("windows/arm64", "windows/arm64/v8") do not yet exist for windows.
- we don't (yet) include `os-variant` for completion (as can be used for Windows images)
- we don't (yet) include platforms for which we don't build binaries, such as
BSD platforms (freebsd, netbsd, openbsd), android, macOS (darwin).
- we currently exclude architectures that may have unofficial builds,
but don't have wide adoption (and no support), such as loong64, mipsXXX,
ppc64 (non-le) to prevent confusion.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This commit adds tests for the commands
docker kill, docker commit, and docker
pause. Also, it creates the mock methods
of the docker client ContainerCommit and
ContainerPause so they can
be used in the tests.
For docker kill, it covers the
cases that:
- the command runs successfully
- the client returns an error
For docker commit, it covers
the cases that:
- the command runs successfully
- the client returns an error
For docker pause, it covers
the cases that:
- the command runs successfully
- the client returns an error
Signed-off-by: Stavros Panakakis <stavrospanakakis@gmail.com>
Move the code for parsing key-value files, such as used for
env-files and label-files to a separate package. This allows
other projects (such as compose) to use the same parsing
logic, but provide custom lookup functions for their situation
(which is slightly different).
The new package provides utilities for parsing key-value files
for either a file or an io.Reader. Most tests for EnvFile were
now testing functionality that's already tested in the new package,
so were (re)moved.
Co-authored-by: Nicolas De Loof <nicolas.deloof@gmail.com>
Signed-off-by: Nicolas De Loof <nicolas.deloof@gmail.com>
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
We used a hard-coded list of capabilities that we copied from containerd,
but the new "capability" package allows use to have a maintained list
of capabilities.
There's likely still some improvements to be made;
First of all, the capability package could provide a function to get the list
of strings.
On the completion-side, we need to consider what format is most convenient;
currently we use the canonical name (uppercase and "CAP_" prefix), however,
tab-completion is case-sensitive by default, so requires the user to type
uppercase letters to filter the list of options.
Bash completion provides a `completion-ignore-case on` option to make completion
case-insensitive (https://askubuntu.com/a/87066), but it looks to be a global
option; the current cobra.CompletionOptions also don't provide this as an option
to be used in the generated completion-script.
Fish completion has `smartcase` (by default?) which matches any case if
all of the input is lowercase.
Zsh does not have a dedicated option, but allows setting matching-rules
(see https://superuser.com/a/1092328).
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This renames the `--time` flag as used on `docker stop` and `docker restart`
to `--timeout`, bringing it in line with other uses for this property,
such as `--stop-timeout` on `docker run`.
The `--time` option is deprecated and hidden, but will be kept for
backward compatibility, as these options existed for a long time.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
full diff: 2269acc7a3...164cae56ed
Co-authored-by: Paweł Gronowski <pawel.gronowski@docker.com>
Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This commit adds tests for the commands
docker diff and docker rename. Also,
it creates the mock methods of the
docker client ContainerDiff and
ContainerRename so they can
be used in the tests.
For docker diff, it covers the
cases that:
- the command runs successfully
- the client returns an error
- the container id is empty
For docker rename, it covers
the cases that:
- the command runs successfully
- the container old name is empty
- the container new name is empty
- the client returns an error
Co-authored-by: Laura Brehm <laurabrehm@hey.com>
Signed-off-by: Stavros Panakakis <stavrospanakakis@gmail.com>
Commit 964155cd tried to enclose all IPv6 addresses within brackets but
missed some cases. This commit fixes that, and adds a few test cases.
Signed-off-by: Albin Kerouanton <albinker@gmail.com>
On Windows, the drive casing doesn't matter outside of WSL. For WSL, the
drives are lowercase. When we're producing a WSL path, lowercase the
drive letter.
Co-authored-by: Jonathan A. Sternberg <jonathan.sternberg@docker.com>
Co-authored-by: Laura Brehm <laurabrehm@hey.com>
Signed-off-by: Jonathan A. Sternberg <jonathan.sternberg@docker.com>
Signed-off-by: Laura Brehm <laurabrehm@hey.com>
This checks for the equivalent WSL mount path on windows. WSL will mount
the windows drives at `/mnt/c` (or whichever drive is being used).
This is done by parsing a UNC path with forward slashes from the unix
socket URL.
Signed-off-by: Jonathan A. Sternberg <jonathan.sternberg@docker.com>
The `Commit` type was introduced in 2790ac68b3,
to assist triaging issues that were reported with an incorrect version of
runc or containerd. At the time, both `runc` and `containerd` were not yet
stable, and had to be built from a specific commit to guarantee compatibility.
We encountered various situations where unexpected (and incompatible) versions
of those binaries were packaged, resulting in hard to trace bug-reports.
For those situations, a "expected" version was set at compile time, to
indicate if the version installed was different from the expected version;
docker info
...
runc version: a592beb5bc4c4092b1b1bac971afed27687340c5 (expected: 69663f0bd4b60df09991c08812a60108003fa340)
Both `runc` and `containerd` are stable now, and docker 19.03 and up set the
expected version to the actual version since c65f0bd13c
and 23.0 did the same for the `init` binary b585c64e2b,
to prevent the CLI from reporting "unexpected version".
In short; the `Expected` fields no longer serves a real purpose, so we should
no longer print it.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This command was declaring that it requires at least 1 argument, when it
needs exactly 1 argument. This was causing the CLI to panic when the
command was invoked with no argument:
`docker volume update`
Signed-off-by: Laura Brehm <laurabrehm@hey.com>
Running `docker login` in a non-interactive environment sometimes errors
out if no username/pwd is provided. This handling is somewhat
inconsistent – this commit addresses that.
Before:
| `--username` | `--password` | Result |
|:------------:|:------------:| ------------------------------------------------------------------ |
| ✅ | ✅ | ✅ |
| ❌ | ❌ | `Error: Cannot perform an interactive login from a non TTY device` |
| ✅ | ❌ | `Error: Cannot perform an interactive login from a non TTY device` |
| ❌ | ✅ | hangs |
After:
| `--username` | `--password` | Result |
|:------------:|:------------:| ------------------------------------------------------------------ |
| ✅ | ✅ | ✅ |
| ❌ | ❌ | `Error: Cannot perform an interactive login from a non TTY device` |
| ✅ | ❌ | `Error: Cannot perform an interactive login from a non TTY device` |
| ❌ | ✅ | `Error: Cannot perform an interactive login from a non TTY device` |
It's worth calling out a separate scenario – if there are previous,
valid credentials, then running `docker login` with no username or
password provided will use the previously stored credentials, and not
error out.
```console
cat ~/.docker/config.json
{
"auths": {
"https://index.docker.io/v1/": {
"auth": "xxxxxxxxxxx"
}
}
}
⭑ docker login 0>/dev/null
Authenticating with existing credentials...
Login Succeeded
```
This commit also applies the same non-interactive handling logic to the
new web-based login flow, which means that now, if there are no prior
credentials stored and a user runs `docker login`, instead of initiating
the new web-based login flow, an error is returned.
Signed-off-by: Laura Brehm <laurabrehm@hey.com>
Previously, if while polling for oauth device-code login results a user
suspended the process (such as with CTRL-Z) and then restored it with
`fg`, an error might occur in the form of:
```
failed waiting for authentication: You are polling faster than the specified interval of 5 seconds.
```
This is due to our use of a `time.Ticker` here - if no receiver drains
the ticker channel (and timers/tickers use a buffered channel behind the
scenes), more than one tick will pile up, causing the program to "tick"
twice, in fast succession, after it is resumed.
The new implementation replaces the `time.Ticker` with a `time.Timer`
(`time.Ticker` is just a nice wrapper) and introduces a helper function
`resetTimer` to ensure that before every `select`, the timer is stopped
and it's channel is drained.
Signed-off-by: Laura Brehm <laurabrehm@hey.com>
Normalization/converting the registry address to just a hostname happens
inside of `command.GetDefaultAuthConfig`. Use this value for the rest of
the login flow/storage.
Signed-off-by: Laura Brehm <laurabrehm@hey.com>
This reverts commit e6624676e0.
Since e6624676e0, during login, we started
normalizing `registry-1.docker.io` to `index.docker.io`. This means that
if a user logs in with `docker login -u [username]
registry-1.docker.io`, the user's credentials get stored in
credhelpers/config.json under `https://index.docker.io/v1/`.
However, while the registry code normalizes an image reference without
registry (`docker pull alpine:latest`) and image references explicitly for
`index.docker.io` (`docker pull index.docker.io/library/alpine:latest`)
to the official index server (`https://index.docker.io/v1/`), and
fetches credentials for that auth key, it does not normalize
`registry-1.docker.io`, which means pulling explicitly from there
(`docker pull registry-1.docker.io/alpine:latest`) will not use
credentials stored under `https://index.docker.io/v1/`.
As such, until changes are made to the registry/pull/push code to
normalize `registry-1.docker.io` to `https://index.docker.io/v1/`, we
should not normalize this during login.
Signed-off-by: Laura Brehm <laurabrehm@hey.com>
cli/required.go:33:22: param min has same name as predeclared identifier (predeclared)
func RequiresMinArgs(min int) cobra.PositionalArgs {
^
cli/required.go:50:22: param max has same name as predeclared identifier (predeclared)
func RequiresMaxArgs(max int) cobra.PositionalArgs {
^
cli/required.go:67:24: param min has same name as predeclared identifier (predeclared)
func RequiresRangeArgs(min int, max int) cobra.PositionalArgs {
^
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
cli/command/utils.go:225:29: printf: non-constant format string in call to github.com/pkg/errors.Wrapf (govet)
return errors.Wrapf(err, fmt.Sprintf("invalid output path: %q must be a directory or a regular file", path))
^
cli/command/manifest/cmd.go:21:33: printf: non-constant format string in call to fmt.Fprintf (govet)
fmt.Fprintf(dockerCli.Err(), "\n"+cmd.UsageString())
^
cli/command/service/remove.go:45:24: printf: non-constant format string in call to github.com/pkg/errors.Errorf (govet)
return errors.Errorf(strings.Join(errs, "\n"))
^
cli/command/service/scale.go:93:23: printf: non-constant format string in call to github.com/pkg/errors.Errorf (govet)
return errors.Errorf(strings.Join(errs, "\n"))
^
cli/command/stack/swarm/remove.go:74:24: printf: non-constant format string in call to github.com/pkg/errors.Errorf (govet)
return errors.Errorf(strings.Join(errs, "\n"))
^
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
cli/command/system/info.go:375:5: S1009: should omit nil check; len() for []github.com/docker/docker/api/types/system.NetworkAddressPool is defined as zero (gosimple)
if info.DefaultAddressPools != nil && len(info.DefaultAddressPools) > 0 {
^
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
On `docker ps`, port bindings with an IPv6 HostIP should have their
addresses put into brackets when joining them to their ports.
RFC 3986 (Section 3.2.2) stipulates that IPv6 addresses should be
enclosed within square brackets. This RFC is only about URIs. However,
doing so here helps user identifier what's part of the IP address and
what's the port. It also makes it easier to copy/paste that
'[addr]:port' into other software (including browsers).
Signed-off-by: Albin Kerouanton <albinker@gmail.com>
Don't output the extra spacing around the images when none of the
top-level image entries has any children.
This makes the list look better when ran against the graphdrivers image
store.
Signed-off-by: Paweł Gronowski <pawel.gronowski@docker.com>
This commit adds support for the oauth [device-code](https://auth0.com/docs/get-started/authentication-and-authorization-flow/device-authorization-flow)
login flow when authenticating against the official registry.
This is achieved by adding `cli/internal/oauth`, which contains code to manage
interacting with the Docker OAuth tenant (`login.docker.com`), including launching
the device-code flow, refreshing access using the refresh-token, and logging out.
The `OAuthManager` introduced here is also made available through the `command.Cli`
interface method `OAuthManager()`.
In order to maintain compatibility with any clients manually accessing
the credentials through `~/.docker/config.json` or via credential
helpers, the added `OAuthManager` uses the retrieved access token to
automatically generate a PAT with Hub, and store that in the
credentials.
Signed-off-by: Laura Brehm <laurabrehm@hey.com>
The addSSHTimeout and disablePseudoTerminalAllocation were added in commits
a5ebe2282a and f3c2c26b10,
and called inside the Dialer function, which means they're called every
time the Dialer is called. Given that the sshFlags slice is not mutated
by the Dialer, we can call these functions once.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
avoided the join, also did manual iteration
added test, also added reflect for the DeepEqual comparison
Signed-off-by: Archimedes Trajano <developer@trajano.net>
This test was just incorrect (and testing incorrect
behavior): it was checking that `docker run` exited with a `context
canceled` error after signalling the CLI/cancelling the command's
context, but this was incorrect (and was fixed in
991b1303da - which was when this test
started failing).
However, since this test assertion was happening inside of a goroutine,
it would sometimes pass if this assertion didn't get to run before the
test suite terminated. It was flaky because sometimes this assertion
inside the goroutine did get to execute, but after the test finished
execution, which is a big no-no.
As an aside, assertions inside goroutines are generally bad, and `govet`
even has a linter for this (but it only catches `t.Fatal` and `t.FailNow`
calls and not `assert.Xx`.
Signed-off-by: Laura Brehm <laurabrehm@hey.com>
Such as with `docker run`, if a user CTRL-Cs while attached to a
container, we should forward the signal and wait for the exit from
`ContainerWait`, instead of just returning.
Signed-off-by: Laura Brehm <laurabrehm@hey.com>
In 3f0d90a2a9 we introduced a global
signal handler and made sure all the contexts passed into command
execution get (appropriately) cancelled when we get a SIGINT.
Due to that change, and how we use this context during `docker attach`,
we started to return the context cancelation error when a user signals
the running `docker attach`.
Since this is the intended behavior, we shouldn't return an error, so
this commit adds checks to ignore this specific error in this case.
Also adds a regression test.
Signed-off-by: Laura Brehm <laurabrehm@hey.com>
Sebastiaan van Stijn
Harald Albers
Paweł Gronowski
Stavros Panakakis
Sebastiaan van Stijn
Nicolas De Loof
Rob Murray
Brian Goff
Djordje Lukic
Albin Kerouanton
Yves Brissaud
Jonathan A. Sternberg
Laura Brehm
lentil32
David Karlsson
Archimedes Trajano