From the Bash manual's `set -e` description:
(https://www.gnu.org/software/bash/manual/bashref.html#index-set)
> Exit immediately if a pipeline (see Pipelines), which may consist of a
> single simple command (see Simple Commands), a list (see Lists), or a
> compound command (see Compound Commands) returns a non-zero status.
> The shell does not exit if the command that fails is part of the
> command list immediately following a while or until keyword, part of
> the test in an if statement, part of any command executed in a && or
> || list except the command following the final && or ||, any command
> in a pipeline but the last, or if the command’s return status is being
> inverted with !. If a compound command other than a subshell returns a
> non-zero status because a command failed while -e was being ignored,
> the shell does not exit.
Additionally, further down:
> If a compound command or shell function executes in a context where -e
> is being ignored, none of the commands executed within the compound
> command or function body will be affected by the -e setting, even if
> -e is set and a command returns a failure status. If a compound
> command or shell function sets -e while executing in a context where
> -e is ignored, that setting will not have any effect until the
> compound command or the command containing the function call
> completes.
Thus, the only way to have our `.integration-daemon-stop` script
actually run appropriately to clean up our daemon on test/script failure
is to use `trap ... EXIT`, which we traditionally avoid because it does
not have any stacking capabilities, but in this case is a reasonable
compromise because it's going to be the only script using it (for now,
at least; we can evaluate more complex solutions in the future if they
actually become necessary).
The alternatives were much less reasonable. One is to have the entire
complex chains in any script wanting to use `.integration-daemon-start`
/ `.integration-daemon-stop` be chained together with `&&` in an `if`
block, which is untenable. The other I could think of was taking the
body of these scripts out into separate scripts, essentially meaning
we'd need two files for each of these, which further complicates the
maintenance.
Add to that the fact that our `trap ... EXIT` is scoped to the enclosing
subshell (`( ... )`) and we're in even more reasonable territory with
this pattern.
Signed-off-by: Andrew "Tianon" Page <admwiggin@gmail.com>
This also removes the now-defunct `*maintainer*.sh` scripts that don't work with the new TOML format, and moves a couple not-build-or-release-related scripts to `contrib/` instead.
Signed-off-by: Andrew "Tianon" Page <admwiggin@gmail.com>
We might want to break it up into smaller pieces (eg. tools in one
place, documents in another) but let's worry about that later.
Signed-off-by: Solomon Hykes <solomon@docker.com>
Without this patch, containers inherit the open file descriptors of the daemon, so my "exec 42>&2" allows us to "echo >&42 some nasty error with some bad advice" directly into the daemon log. :)
Also, "hack/dind" was already doing this due to issues caused by the inheritance, so I'm removing that hack too since this patch obsoletes it by generalizing it for all containers.
Docker-DCO-1.1-Signed-off-by: Andrew Page <admwiggin@gmail.com> (github: tianon)
Creating the "docker save" tarball for "scratch" is pretty simple. I've also extrapolated the "docker build -t busybox ." logic into a separate "hack/make/.ensure-busybox" file so that it can eventually be reused easier.
Docker-DCO-1.1-Signed-off-by: Andrew Page <admwiggin@gmail.com> (github: tianon)
Don't make calls to the registry if the image exists already.
Docker-DCO-1.1-Signed-off-by: Cristian Staretu <cristian.staretu@gmail.com> (github: unclejack)
"set -e" is already inherited here from make.sh, but explicit is always better than implicit (hence the "set -e" in the first place!)
Docker-DCO-1.1-Signed-off-by: Andrew Page <admwiggin@gmail.com> (github: tianon)
The wait at the end of cli integration script could end up failing if
the process had already exited. This was making it look like the tests
have failed.
This change fixes the problem.
Docker-DCO-1.1-Signed-off-by: Cristian Staretu <cristian.staretu@gmail.com> (github: unclejack)
This makes it possible to choose the graphdriver and the execdriver
which is going to be used for the cli integration tests.
Docker-DCO-1.1-Signed-off-by: Cristian Staretu <cristian.staretu@gmail.com> (github: unclejack)
Make sure the busybox image is ready to be used when running the cli
integration tests.
Docker-DCO-1.1-Signed-off-by: Cristian Staretu <cristian.staretu@gmail.com> (github: unclejack)
Please do with this as you please (including rebasing and/or squashing it), especially under clause (c) of the DCO.
Docker-DCO-1.1-Signed-off-by: Andrew Page <admwiggin@gmail.com> (github: tianon)
