Starting with this commit, integration tests should no longer rely on
the docker cli, they should be API tests instead. For the existing tests
the scripts will use a frozen version of the docker cli with a
DOCKER_API_VERSION frozen to 1.30, which should ensure that the CI remains
green at all times.
To help contributors develop and test manually with a modified docker
cli, this commit also adds a DOCKER_CLI_PATH environment variable to the
Makefile. This allows to set the path of a custom cli that will be
available inside the development container and used to run the
integration tests.
Signed-off-by: Arnaud Porterie (icecrime) <arnaud.porterie@docker.com>
Signed-off-by: Tibor Vass <tibor@docker.com>
This updates the versions of the frozen images used to their current
version. The original reason for updating these images was to make sure
they are not affected by [CVE-2016-1252 / DSA-3733-1](https://lwn.net/Articles/709119/),
which is fixed in apt 1.0.9.8.4 and up.
Note that `CVE-2016-1252` won't affect our test-suite, because no packages
are installed during out tests. It is just "good practice" to keep these
images up to date.
The `debian:jessie`, and `buildpack-deps:jessie` in `Dockerfile.s390x`,
and `Dockerfile.armhf` have not been updated in this patch, because
those images have not yet been updated to contain apt 1.0.9.8.4.
While working on this, the `busybox` and `hello-world` were also updated
to their latest version.
Also removes a reference to `hack/make/.ensure-frozen-images`, which
was removed in ff91276d1f.
The new busybox image has one layer less than the original,
so updated `TestBuildSquashParent` to take that into account.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Running the rm command on a paused/restarting container
will give an error message saying the container is running
which is incorrect.
To fix that, the error message will have the correct
container state and a procedure to remove it accordingly.
Notice: docker-py was bumped to:
4a08d04aef0595322e1b5ac7c52f28a931da85a5
Signed-off-by: Boaz Shuster <ripcurld.github@gmail.com>
This drops support for migrations from pre-1.10 Docker versions, which
should be done via an external tool or an intermediate upgrade.
Signed-off-by: Justin Cormack <justin.cormack@docker.com>
The keyserver infrastructure is unreliable, and just adds another point
of failure without adding any security. Instead, commit the key used at
build time for ZFS to the repo, and inline our signing key into the
install script rather than just its fingerprint.
fix#28510fix#13555
Signed-off-by: Justin Cormack <justin.cormack@docker.com>
criu 2.8 and 2.9 contain various fixes,
so updating the version in the Dockerfile
to match the newer version.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
Why? Most of the time I end up needing an editor when in `make shell`.
Spent much time doing `apt-get update && apt-get install vim`.
Since we're already installing vim-common anyway...
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
There is no reason to duplicate efforts and tini is well built and
better than grimes. It is a much stronger option for the default init
and @krallin has done a great job maintaining it and helping make
changes so that it will work with Docker.
Signed-off-by: Michael Crosby <crosbymichael@gmail.com>
- yamllint to ensure it is a valid YAML file
- go-swagger validate to ensure it is a valid swagger file
Signed-off-by: Ben Firshman <ben@firshman.co.uk>
this switches all go download links to use a redirect,
https://golang.org/dl/ instead of the direct source.
Signed-off-by: Christopher Jones <tophj@linux.vnet.ibm.com>
This means we can vendor libnetwork without special casing, and
it is built the same way as the other external binaries.
Signed-off-by: Justin Cormack <justin.cormack@docker.com>
- What I did
While building docker from source, to get the dependencies installed had done make build, then had got this error.
- How I did it
In the DockerFIle, instead using space a tab was put, which when was done make build the next line was getting combined and was unable to install the package.
image
image
Refer the below Hex View of the earlier file.
image
- How to verify it
After fixing, changing tab to space, built from source to install dependencies and was success
- Description for the changelog
Fixing Issue #27035
Signed-off-by: Rojin George itsmerojin@gmail.com
Signed-off-by: rojingeorge <itsmerojin@gmail.com>
This adds a small C binary for fighting zombies. It is mounted under
`/dev/init` and is prepended to the args specified by the user. You
enable it via a daemon flag, `dockerd --init`, as it is disable by
default for backwards compat.
You can also override the daemon option or specify this on a per
container basis with `docker run --init=true|false`.
You can test this by running a process like this as the pid 1 in a
container and see the extra zombie that appears in the container as it
is running.
```c
int main(int argc, char ** argv) {
pid_t pid = fork();
if (pid == 0) {
pid = fork();
if (pid == 0) {
exit(0);
}
sleep(3);
exit(0);
}
printf("got pid %d and exited\n", pid);
sleep(20);
}
```
Signed-off-by: Michael Crosby <crosbymichael@gmail.com>
Interactive integration testing is useful when you're developing new tests, or
making changes to cli code.
Signed-off-by: Daniel Nephin <dnephin@docker.com>
following the announcement;
https://groups.google.com/forum/m/#!topic/golang-announce/7JTsd70ZAT0
> [security] Go 1.6.3 and Go 1.7rc2 pre-announcement
>
> Hello gophers,
> We plan to issue Go 1.6.3 and Go 1.7rc2 on Monday July 18 at approximately 2am UTC.
> These are minor release to fix a security issue.
>
> Following our policy at https://golang.org/security, this is the pre-announcement of those releases.
>
> Because we are so late in the release cycle for Go 1.7, we will not issue a minor release of Go 1.5.
> Additionally, we plan to issue Go 1.7rc3 later next week, which will include any changes between 1.7rc1 and tip.
>
> Cheers,
> Chris on behalf of the Go team
**Note:**
the man/Dockerfile is not yet updated, because
the official image for Go 1.6.2 has not yet
been updated.
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This version introduces the following:
- uses nanosecond timestamps for event
- ensure events are sent once their effect is "live"
Signed-off-by: Kenfe-Mickael Laventure <mickael.laventure@gmail.com>
This updates containerd to b93a33be39bc4ef0fb00bfcb79147a28c33d9d43
fixing the start sync issues.
Signed-off-by: Michael Crosby <crosbymichael@gmail.com>
Due to the issue of go-md2man, a numbered list in `man docker login` was not rendered correctly.
a8f937e113
Signed-off-by: Akihiro Suda <suda.akihiro@lab.ntt.co.jp>
This bumps containerd to cf554d59dd96e459544748290eb9167f4bcde509 and
includes various fixes and updates the grpc package and types generated
for use.
Signed-off-by: Michael Crosby <crosbymichael@gmail.com>
the llvm apt repo was removed and finding the exact rev of the llvm
toolchain that worked perfecting in our dockerfile is a problem.
We looked at the reasons why we were using this version of clang and it
appears to be this compiler warning.
`was built for newer OSX version (10.11) than being linked
(10.6)`
When you look at the dockerfile this makes sense.
```
ENV OSX_SDK MacOSX10.11.sdk
ENV OSX_CROSS_COMMIT 8aa9b71a394905e6c5f4b59e2b97b87a004658a4
RUN set -x \
&& export OSXCROSS_PATH="/osxcross" \
&& git clone https://github.com/tpoechtrager/osxcross.git $OSXCROSS_PATH \
&& ( cd $OSXCROSS_PATH && git checkout -q $OSX_CROSS_COMMIT) \
&& curl -sSL https://s3.dockerproject.org/darwin/v2/${OSX_SDK}.tar.xz -o "${OSXCROSS_PATH}/tarballs/${OSX_SDK}.tar.xz" \
&& UNATTENDED=yes OSX_VERSION_MIN=10.6 ${OSXCROSS_PATH}/build.sh
ENV PATH /osxcross/target/bin:$PATH
```
We are basically using the 10.11 sdk but linking to 10.6 as the
min version so this warning should be expected.
Also the docs on the osxcross project require clan 3.2+, not 3.8.
Signed-off-by: Michael Crosby <crosbymichael@gmail.com>
This environment variable is no longer
needed in Go 1.6 (as it's not the default).
Removed this environment variable from
all Dockerfiles except the Dockerfile.s390x,
which is still using gcc 5.3 (Go 1.5)
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
This change adds file version information to docker.exe and dockerd.exe by
adding a Windows version resource with the windres tool.
This change adds a dependency to binutils-mingw-w64 on Linux, but removes
a dependency on rsrc. Most Windows build environments should already have
windres if they have gcc (which is necessary to build dockerd).
Signed-off-by: John Starks <jostarks@microsoft.com>
For context: https://github.com/golang/go/issues/15286
This commit downloads go1.5.3 in addition to go1.5.4 in order to
workaround the issue.
It is not expected to do a Docker release without a proper fix, however
this should help unblock Docker development on Windows TP5.
Signed-off-by: Tibor Vass <tibor@docker.com>
This vendors in new spec/runc that supports
setting readonly and masked paths in the
configuration. Using this allows us to make an
exception for `—-privileged`.
Signed-off-by: Tonis Tiigi <tonistiigi@gmail.com>
This includes fixes for;
- outputing errors for missing seccomp options on seccomp versions < 2.3
- cap set apply EPERM errors on ARM systems
Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
runc expects a systemd cgroupsPath to be in slice:scopePrefix:containerName
format and the "--systemd-cgroup" option to be set. Update docker accordingly.
Fixes 21475
Signed-off-by: Anusha Ragunathan <anusha@docker.com>
Add some missing dependencies in the Dockerfile:
- apt-utils for apt-ftparchive.conf
- bsdmainutils for our use of the column command in
hack/make/generate-index-listing
We also ensure that the docker daemon is started before calling
release-deb or release-rpm, since .detect-daemon-osarch, which is sourced
in each of them, requires the daemon to be running.
This commit also gets completely rid of s3cmd and fixes references to
AWS_* environment variables (changing from AWS_ACCESS_KEY to
AWS_ACCESS_KEY_ID and AWS_SECRET_KEY to AWS_SECRET_ACCESS_KEY) in order
to please awscli. Also AWS_DEFAULT_REGION is now important to specify,
the default has been set to the region used by get.docker.com and
test.docker.com.
Signed-off-by: Tibor Vass <tibor@docker.com>
Contains fixes for:
- pid.max fix that is causing hang on network stats test.
- fix for early stdin close containerd-shim
- better logging for `could not synchronise with container process`
Signed-off-by: Tonis Tiigi <tonistiigi@gmail.com>
Fixes#20550
This update to libseccomp supports the new versions of socket
system calls that can be called directly rather than via the
socketcall syscall in kernel versions 4.3 or later with new glibc.
Note this library version now supports s390x and ppc64le, so
seccomp can be potentially be enabled for these architectures now.
Signed-off-by: Justin Cormack <justin.cormack@docker.com>
Fixes broken-pipe issue when piping s3cmd to grep -q, by removing the -q
flag and redirecting to /dev/null instead.
Add net-tools for ifconfig, because some tests rely on ifconfig.
Harmonize all Dockerfiles in this direction.
Signed-off-by: Tibor Vass <tibor@docker.com>
dockerinit has been around for a very long time. It was originally used
as a way for us to do configuration for LXC containers once the
container had started. LXC is no longer supported, and /.dockerinit has
been dead code for quite a while. This removes all code and references
in code to dockerinit.
Signed-off-by: Aleksa Sarai <asarai@suse.com>
Fixes#19400
Note that this introduces an incompatibility with Docker 1.10-rc1,
because the media type used for schema1 manifests has been corrected in
the upstream distribution code. Docker 1.10-rc1 won't be able to pull
old manifests from Registry 2.3-rc0 and up, but because of this vendor
update, Docker 1.10-rc2 won't have this problem.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
This results in a significant time savings during repeated builds (since we don't have to re-download gcc for every test run).
Signed-off-by: Andrew "Tianon" Page <admwiggin@gmail.com>
Tracks source repository information for each blob in the blobsum
service, which is then used to attempt to mount blobs from another
repository when pushing instead of having to re-push blobs to the same
registry.
Signed-off-by: Brian Bland <brian.bland@docker.com>
Add DockerSchema1RegistrySuite which uses this registry, and make
applicable integration tests run as part of this suite.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Also, fix up some minor whitespace consistency issues, remove a little cruft, and update GOARM for armhf to 7 so that we're explicit.
Signed-off-by: Andrew "Tianon" Page <admwiggin@gmail.com>
It will Tar up contents of child directory onto tmpfs if mounted over
This patch will use the new PreMount and PostMount hooks to "tar"
up the contents of the base image on top of tmpfs mount points.
Signed-off-by: Dan Walsh <dwalsh@redhat.com>
devmapper uses xfs by default now. So include xfsprogs in build
environment. Also update docs to reflect the new default.
Signed-off-by: Anusha Ragunathan <anusha@docker.com>
The LXC driver was deprecated in Docker 1.8.
Following the deprecation rules, we can remove a deprecated feature
after two major releases. LXC won't be supported anymore starting on Docker 1.10.
Signed-off-by: David Calavera <david.calavera@gmail.com>
Also requires some tests to be updated which relied on behavior
of a busybox image that wasn't actually "busybox:latest"; meaning these
tests were unable to be verified/run against a real busybox:latest image
on a daemon.
Docker-DCO-1.1-Signed-off-by: Phil Estes <estesp@linux.vnet.ibm.com> (github: estesp)
If a logdriver doesn't register a callback function to validate log
options, it won't be usable. Fix the journald driver by adding a dummy
validator.
Teach the client and the daemon's "logs" logic that the server can also
supply "logs" data via the "journald" driver. Update documentation and
tests that depend on error messages.
Add support for reading log data from the systemd journal to the
journald log driver. The internal logic uses a goroutine to scan the
journal for matching entries after any specified cutoff time, formats
the messages from those entries as JSONLog messages, and stuffs the
results down a pipe whose reading end we hand back to the caller.
If we are missing any of the 'linux', 'cgo', or 'journald' build tags,
however, we don't implement a reader, so the 'logs' endpoint will still
return an error.
Make the necessary changes to the build setup to ensure that support for
reading container logs from the systemd journal is built.
Rename the Jmap member of the journald logdriver's struct to "vars" to
make it non-public, and to make it easier to tell that it's just there
to hold additional variable values that we want journald to record along
with log data that we're sending to it.
In the client, don't assume that we know which logdrivers the server
implements, and remove the check that looks at the server. It's
redundant because the server already knows, and the check also makes
using older clients with newer servers (which may have new logdrivers in
them) unnecessarily hard.
When we try to "logs" and have to report that the container's logdriver
doesn't support reading, send the error message through the
might-be-a-multiplexer so that clients which are expecting multiplexed
data will be able to properly display the error, instead of tripping
over the data and printing a less helpful "Unrecognized input header"
error.
Signed-off-by: Nalin Dahyabhai <nalin@redhat.com> (github: nalind)
Added notary server to docker base image.
Created trust suite which runs trust server for running trusted commands.
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
The docker-py commit used in the standard `Dockerfile` is from Feb. 2015
and is out of date with the current API level and has fixes for things
like the new docker cli config location and registry v2 changes/API
responses as well.
Also pass "NOT_ON_HOST=true" to docker-py test suite so that tests
relying on direct HOST interaction (versus running in a container) are
skipped.
Docker-DCO-1.1-Signed-off-by: Phil Estes <estesp@linux.vnet.ibm.com> (github: estesp)
Add golint to the Dockerfile, and a `validate-lint` task to the
Makefile. Currently, the linter will process a harcoded list of packages
that will expand as we fix more warnings. Eventually, the linter should
process all subpackages of the repo (excluding vendored code).
Signed-off-by: Arnaud Porterie <arnaud.porterie@docker.com>
Fixes an issue with curly braces being escaped when they should not be.
This was particularly an issue in places where `--format '{{ ... }}'`
is used in the man docs.
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
These will create the apt & yum repos for the deb/rpms generated by build-deb
and build-rpm.
Adds sign-repo script which signs the repo metadata with a gpg key.
Signed-off-by: Jessica Frazelle <princess@docker.com>
These have both been updated to no longer include the empty "scratch" layer, and `hello-world` has a fix for http->https.
Signed-off-by: Andrew "Tianon" Page <admwiggin@gmail.com>
We already vendor distribution under ./vendor, but
because the GOPATH is /go:/go/src/github.com/.../vendor
Go will always compile the source code at /go not in ./vendor.
Apart from the fact that it is very inconvenient during
development, it was also a time-bomb: someone vendors a fix
from upstream distribution, but forgets to update
REGISTRY_COMMIT in the Dockerfile, and the binary doesn't get
the fix.
Signed-off-by: Tibor Vass <tibor@docker.com>
When ever something vendor/ changes the go dependencies have to downloaded again,
which requires internet access and there for is potential slow. COPY and go install is much faster, while the git urls does not change not this often.
Signed-off-by: Jörg Thalheim <joerg@higgsboson.tk>
Installs and configures bash completion for Docker.
Note that bash completion still has to be initialized by a custom
.bashrc file.
Signed-off-by: Harald Albers <github@albersweb.de>
This addresses a subtle deadlock where an error during a copy prevented pipe
closure to propagate correctly. By closing down the read end of the pipe rather
than the write end, the waiting writer is properly signaled. A nice side-effect
of this change is that errors encountered by io.Copy are now propagated to the
verifier's Write method.
A test to ensure validation errors for unsupported digest types has been added,
as well.
Signed-off-by: Stephen J Day <stephen.day@docker.com>
Add ability to refer to an image by repository name and digest using the
format repository@digest. Works for pull, push, run, build, and rmi.
Signed-off-by: Andy Goldstein <agoldste@redhat.com>
Update pull code to consider any layer download or new tag as an update.
Update hello-world frozen image to be explicitly tagged as frozen, to not interfere with pull tests. The hello-world is used by pull tests because of its small size and there is no other official image with such a size.
fixes#11383
Signed-off-by: Derek McGowan <derek@mcgstyle.net> (github: dmcgowan)
FROM scratch
ADD echo /
RUN [ "echo" ]
will die with
exec: "/bin/sh": stat /bin/sh: no such file or directory
Signed-off-by: Doug Davis <dug@us.ibm.com>
This makes it much simpler to add new "frozen" images -- simply add them to the `Dockerfile` and in `hack/make/.ensure-frozen-images` and you're off to the races.
Signed-off-by: Andrew "Tianon" Page <admwiggin@gmail.com>
This downloads a specific image ID of `busybox:latest` from the Hub directly (within the `Dockerfile`, ready for `docker load`) instead of grabbing the source from GitHub and doing a `docker build` at daemon start time. This ensures the test suite runs more consistently.
Signed-off-by: Andrew "Tianon" Page <admwiggin@gmail.com>
If `.bashrc` exists at the root of the source tree, it will be used as the `~/.bashrc` inside the container.
Signed-off-by: Andrew "Tianon" Page <admwiggin@gmail.com>
Since `cirros.tar.gz` only existed to test `docker import`'s display and
presence in `docker events`, we can instead just use `docker export`
piped directly to `docker import` to achieve the same goal without
another external dependency besides `busybox` (which we already have).
While I was at it, I updated `TestImportDisplay` to also test that the
imported image actually runs successfully as well (so we're testing the
full import round-trip).
Signed-off-by: Andrew "Tianon" Page <admwiggin@gmail.com>
Use `env -i` to very explicitly control exactly which environment variables leak into our tests. This enforces a clean separation of "build environment knobs" versus "test suite knobs".
This also includes a minor tweak to how we handle starting our integration daemon, especially to catch failure to start sooner than failing tests.
Signed-off-by: Andrew "Tianon" Page <admwiggin@gmail.com>
Update fixes some rendering issues, including improperly escaping '$' in
blocks, and actual parsing of blockcode.
`ID=$(sudo docker run -d fedora /usr/bin/top -b)` was being converted to
`ID=do docker run -d fedora/usr/bin/top -b)`
Signed-off-by: Brian Goff <cpuguy83@gmail.com>
Since it can still be tested natively without this, this won't cause any harm while we fix the tests to actually work on Windows.
Signed-off-by: Andrew "Tianon" Page <admwiggin@gmail.com>
Update our "registry" install to use the included Godeps libraries so that it
doesn't require anything from our current source (hence moving it up for
better caching too)
Signed-off-by: Andrew "Tianon" Page <admwiggin@gmail.com>
v2 ping now checks for a Docker-Distribution-API-Version
header that identifies the endpoint as "registry/2.0"
Docker-DCO-1.1-Signed-off-by: Josh Hawn <josh.hawn@docker.com> (github: jlhawn)
Since "go test" doesn't seem to support "-installsuffix" as quite the same perfect solution that "go build" is happy to let it be, let's just switch those crappy old "integration/" tests to use our separate static dockerinit binary so we don't have to worry about compiling the entire test harness statically. 👍
Signed-off-by: Andrew "Tianon" Page <admwiggin@gmail.com>
This refactors version number usage for several of our "download/install this exact version" bits so we have an easier time replacing the version numbers in Jenkins.
Signed-off-by: Andrew "Tianon" Page <admwiggin@gmail.com>
- move docker/docker-py clone to the Dockerfile
- put "integration test daemon startup" code in a separate file for both scripts to source
- add new test-docker-py Makefile target
- include "python-websocket" package in Dockerfile for running the tests
Signed-off-by: Andrew "Tianon" Page <admwiggin@gmail.com>
This clones and run the integration tests for docker-py master as part
of the integration tests created on master. docker-py hits the api
directly and should be a good way to identify regressions in the api.
Signed-off-by: Michael Crosby <crosbymichael@gmail.com>
Since the build uses ubuntu 14.04, which has an old btrfs, include the
buildtags needed for this old version to not break the build.
Signed-off-by: Vincent Batts <vbatts@redhat.com>
So far, it looks like the declarations are not used, and so its safer not to
confuse people into thinking they do something.
Docker-DCO-1.1-Signed-off-by: Sven Dowideit <SvenDowideit@docker.com> (github: SvenDowideit)
Because of the base image change, $HOME is not always / and we need to
write to the proper $HOME within the container to complete the release
process.
Signed-off-by: Michael Crosby <crosbymichael@gmail.com>
This changes the test TestBuildWithInaccessibleFilesInContext to not
add the user 'unprivilegeduser' and add it via the Dockerfile instead.
Docker-DCO-1.1-Signed-off-by: Cristian Staretu <cristian.staretu@gmail.com> (github: unclejack)
This patch just revs the Ubuntu release used for builds from 13.10
to 14.04.
Docker-DCO-1.1-Signed-off-by: Aleksa Sarai <cyphar@cyphar.com> (github: cyphar)
This also removes all the old man pages, .gitignores their directory, and updates the md2man-all.sh script to be easier to read and more friendly to being called within hack/make/ubuntu.
Docker-DCO-1.1-Signed-off-by: Andrew Page <admwiggin@gmail.com> (github: tianon)
docker will run the process(es) within the container with an SELinux label and will label
all of the content within the container with mount label. Any temporary file systems
created within the container need to be mounted with the same mount label.
The user can override the process label by specifying
-Z With a string of space separated options.
-Z "user=unconfined_u role=unconfined_r type=unconfined_t level=s0"
Would cause the process label to run with unconfined_u:unconfined_r:unconfined_t:s0"
By default the processes will run execute within the container as svirt_lxc_net_t.
All of the content in the container as svirt_sandbox_file_t.
The process mcs level is based of the PID of the docker process that is creating the container.
If you run the container in --priv mode, the labeling will be disabled.
Docker-DCO-1.1-Signed-off-by: Dan Walsh <dwalsh@redhat.com> (github: rhatdan)
These were found using `git grep -nE '[^-a-zA-Z0-9<>]-[a-zA-Z0-9]{2}'` (fair warning: _many_ false positives there).
Docker-DCO-1.1-Signed-off-by: Andrew Page <admwiggin@gmail.com> (github: tianon)
Everyone probably wants to run `docker pull ubuntu` before building with this change.
Docker-DCO-1.1-Signed-off-by: Andrew Page <admwiggin@gmail.com> (github: tianon)
This fixes compilation issues when btrfs.h isn't available (because we just need the relevant structs, which for userspace programs are supposed to come from btrfs-progs instead of the kernel headers).
Docker-DCO-1.1-Signed-off-by: Andrew Page <admwiggin@gmail.com> (github: tianon)
In addition, we've now hard-coded LXC version 0.8 compiled from source so that we can have the most stable dev environment possible.
Docker-DCO-1.1-Signed-off-by: Andrew Page <admwiggin@gmail.com> (github: tianon)
