Helm had been exposing XDG based variables to end users. This lead
to confusion. For example, if a user wanted to change the cache
location Helm used should they change the XDG variable? Since this
would be like changing the HOME environment variable the answer
is no.
This change adds HELM_*_HOME environment variables to be used
in addition to XDG ones of the same name. Helm will now look
for the Helm specific variable. If not set, Helm will fall
back to XDG locations. If those are not set a default location
will be used. This keeps XDG in use as a default when present,
provides users with the ability to set the location, and removes
XDG from being exposed to end users to avoid confusion.
Closes#7919
Signed-off-by: Matt Farina <matt@mattfarina.com>
* fix: write index.yaml file atomically
This refactors the already-existing `AtomicWriteFile` utility
to a central location and uses it to write index files
atomically.
This is done to avoid having half-written index files break
client requests.
Drive-bys:
- Add test for AtomicWriteFile.
- Add test IndexFile.WriteFile.
Signed-off-by: rabadin <rvbadin@gmail.com>
* Review fix: use RenameWithFallback instead of os.Rename
Signed-off-by: rabadin <rvbadin@gmail.com>
Co-authored-by: rabadin <rvbadin@gmail.com>
Removes the BOM prefix if present, in read files before
processing the data.
Affects the following pkg:
- pkg/chart/loader: directory and archive loader
- internal/ignore: when loading .helmignore file
Signed-off-by: Thomas FREYSS <thomas.freyss@gmail.com>
This change lower the expectations on test env setup, i.e. tests could be executed in a container under a random UID,
without require an user in /etc/passwd
Signed-off-by: Predrag Knezevic <pknezevi@redhat.com>
Upgrade Kubernetes libraries to v0.18.0
Add new lazy load KubernetesClientSet to avoid missing kubeconfig error
In kubernetes v1.18 kubeconfig validation was added. Minikube and Kind
both remove kubeconfig when stopping clusters. This causes and error
when running any helm commands because we initialize the client before
executing the command.
Signed-off-by: Adam Reese <adam@reese.io>
* fix: add new static linter and fix issues it found
Signed-off-by: Matt Butcher <matt.butcher@microsoft.com>
* fixed two additional linter errors.
Signed-off-by: Matt Butcher <matt.butcher@microsoft.com>
For each plugin, helm registers a ValidArgsFunction which the completion
script will call when it needs dynamic completion. This
ValidArgsFunction will setup the plugin environment and then call the
executable `plugin.complete`, if it is provided by the plugin.
The output of the call to `plugin.complete` will be used as completion
choices. The last line of the output can optionally be used by the
plugin to specify a completion directive.
Signed-off-by: Marc Khouzam <marc.khouzam@montreal.ca>
* Fix issue with apiVersion v1 lock digest
When apiVersion v1 chart dependencies are built with Helm 2
and then built with Helm 3, the lock digests differ. To avoid
this issue, a depdendency update is forced.
Signed-off-by: Martin Hickey <martin.hickey@ie.ibm.com>
* Check against Helm v2 hash
Handle scenario where dependency hash was generated by Helm v2
but need to do a dependency build with Helm v3.
Signed-off-by: Martin Hickey <martin.hickey@ie.ibm.com>
* Add unit test
Signed-off-by: Martin Hickey <martin.hickey@ie.ibm.com>
* Refactor unit test
Refactor unit test to use an existing chart as dependency
Signed-off-by: Martin Hickey <martin.hickey@ie.ibm.com>
* Update after review
Comments:
- https://github.com/helm/helm/pull/7261#discussion_r373827088
- https://github.com/helm/helm/pull/7261#discussion_r373827250
Signed-off-by: Martin Hickey <martin.hickey@ie.ibm.com>
This newly added tests was failing on MacOS because /proc does not
exist. This commit replaces /proc with /tmp to achieve the same result.
Signed-off-by: Marc Khouzam <marc.khouzam@montreal.ca>
To use error codes to indicate completion directive to the completion
script had us use os.Exit() in the __complete command. This prevented
go tests calling the __complete command from succeeding.
Another option was to return an error containing an error code
(like is done for helm plugins) instead of calling os.Exit(). However
such an approach requires a change in how helm handles the returned
error of a Cobra command; although we can do this for Helm, it would
be an annoying requirement for other programs if we ever push this
completion logic into Cobra.
The chosen solution instead is to printout the directive at the end
of the list of completions, and have the completion script extract it.
Note that we print both the completions and directive to stdout.
It would have been interesting to print the completions to stdout and
the directive to stderr; however, it is very complicated for the
completion script to extract both stdout and stderr to different
variables, and even if possible, such code would not be portable to
many shells. Printing both to stdout is much simpler.
Signed-off-by: Marc Khouzam <marc.khouzam@montreal.ca>
To make digests include information about Chart.yaml dependencies, not
only the lock file, digest calculation is changed to accept both
contents. This terminates the `dep build` command if Chart.yaml
dependencies have been updated so that `dep up` should be executed
properly, to prevent downloading wrong versions or mismatched subcharts.
Note that previous Helm cannot know whether Chart.yaml dependencies were changed
or not since the Chart.lock's digest is calculated by only Chart.lock
contents, which don't include information about SemVer ranges and extra
dependency fields such as aliases, conditions, and tags. Specially,
SemVer can be written as a version range in Chart.yaml, but Chart.lock
has the specific, resolved version of that range.
Signed-off-by: Hang Park <hangpark@kaist.ac.kr>
Mediatype changed to application/tar+gzip. Please see the
following OCI mailing list item for more info:
https://groups.google.com/a/opencontainers.org/forum/#!topic/dev/pdc1lucm_Ak
Also, improved check for invalid manifests, a nil reference error was
occurring when upgrading from existing cache with old mediatype.
Signed-off-by: Josh Dolitsky <393494+jdolitsky@users.noreply.github.com>
This code was ported over from PR #5038, #6738 which were originally for helm
v2. The code contains functions from golang/dep/internal/fs for renaming files.
Signed-off-by: Yagnesh Mistry <ysh@live.in>
We already had the copystructure library in our dependencies transitively
through sprig. This solves a gob encoding bug that was causing issues with
chart testing
Signed-off-by: Taylor Thomas <taylor.thomas@microsoft.com>
this was partially fixed in #6430 but the fix only
worked for values without nesting. this PR fixes it.
this is done by doing a deep copy of values rather
than a top level keys copy. deep copy ensures
values are not mutated during coalesce()
execution which leads to bugs like #6659
the deep copy code has been copied from:
https://gist.github.com/soroushjp/0ec92102641ddfc3ad5515ca76405f4d
which is in turn inspired by this stackoverflow answer:
http://stackoverflow.com/a/28579297/1366283
Signed-off-by: Karuppiah Natarajan <karuppiah7890@gmail.com>
* Drop dependency on k8s.io/kubernetes
https://github.com/helm/helm/issues/6606
Depending on k8s.io/kubernetes is not recommended by Kubernetes, and
forces dependencies of Helm to also depend on them. We are only using
this dependency in one relatively isolated occurance, which can be
easily copied over rather than depending on the entire Kubernetes.
Copying this code is not very desirable, so if we don't want to have
this duplication we can at least use this PR as a PoC and see if we can
get Kubernetes to publish the controller package as a separate Go module
(see
https://github.com/kubernetes/kubernetes/issues/79384#issuecomment-538740756)
Signed-off-by: John Howard <howardjohn@google.com>
* Move to internal
Signed-off-by: John Howard <howardjohn@google.com>
* Exclude third_party from validate-license.sh
Signed-off-by: John Howard <howardjohn@google.com>