- remove go vet and build as they are now redundant
- replace ./pkg with ./act
- rename the workflow from test to act to distinguish it from the
runner workflow
cleanup and factorization with the runner workflow is left for later
To prepare for a smooth merge in the runner codebase.
- run with --fix for gofumpt and golangci
- manual edits for
- disabling useless package naming warning
- rename variables that had underscore in their name
- remove trailing else at the end of a few functions
Reviewed-on: https://code.forgejo.org/forgejo/act/pulls/206
Reviewed-by: Michael Kriese <michael.kriese@gmx.de>
Co-authored-by: Earl Warren <contact@earl-warren.org>
Co-committed-by: Earl Warren <contact@earl-warren.org>
Only changes test files.
---
- remove tests specific to running on a host with no container
they are the same as with the containers (TestRunEventHostEnvironment)
- prefix the name of the tests with a distinctive name to
more easily run them together
- use code.forgejo.org/oci images whereever possible
- remove some tests that are either
- difficult to understand (ancient bugs)
- not yet well understood (related to reusable workflows)
- depend on github (generation of workflows to be run on the fly
e.g. updateTestIfWorkflow)
- fix the TestSanitizeNetworkAlias tests that were not run
and make them easier to debug
- disable tests of options that are forbidden in Forgejo Actions
(testdata/container-hostname/push.yml)
Reviewed-on: https://code.forgejo.org/forgejo/act/pulls/202
Reviewed-by: Michael Kriese <michael.kriese@gmx.de>
Co-authored-by: Earl Warren <contact@earl-warren.org>
Co-committed-by: Earl Warren <contact@earl-warren.org>
- prefer the lowercase version of the DNS name which is more common they are case insensitive anyway
- fix the inverted information when sanitation happens
Reviewed-on: https://code.forgejo.org/forgejo/act/pulls/197
Reviewed-by: msrd0 <msrd0@noreply.code.forgejo.org>
Co-authored-by: Earl Warren <contact@earl-warren.org>
Co-committed-by: Earl Warren <contact@earl-warren.org>
For instance, the volume name derived from the workflow name may exceed the file system limit when the container name it is derived from is too long.
Fixesforgejo/runner#152
Reviewed-on: https://code.forgejo.org/forgejo/act/pulls/191
Reviewed-by: Michael Kriese <michael.kriese@gmx.de>
Co-authored-by: Earl Warren <contact@earl-warren.org>
Co-committed-by: Earl Warren <contact@earl-warren.org>
- s/[^A-Z0-9-]/_/g
- add a log line in case the name is sanitized
Closesforgejo/runner#226
---
It is breaking because it will fail jobs that rely on service names that contain characters that are sanitized
Reviewed-on: https://code.forgejo.org/forgejo/act/pulls/190
Reviewed-by: Michael Kriese <michael.kriese@gmx.de>
Co-authored-by: Earl Warren <contact@earl-warren.org>
Co-committed-by: Earl Warren <contact@earl-warren.org>
In the current implementation, the network will only be created and be scheduled to be deleted when there is a service:
```go
func (rc *RunContext) networkName() (string, bool) {
if len(rc.Run.Job().Services) > 0 || rc.Config.ContainerNetworkMode == "" {
return fmt.Sprintf("%s-%s-network", rc.jobContainerName(), rc.Run.JobID), true
}
return string(rc.Config.ContainerNetworkMode), false
}
```
Therefore it does not currently make a difference. However, in case the network creation logic changes and a network is created even if a service is not present, it would be incorrect not to delete it.
Reviewed-on: https://code.forgejo.org/forgejo/act/pulls/189
Reviewed-by: Michael Kriese <michael.kriese@gmx.de>
Co-authored-by: Earl Warren <contact@earl-warren.org>
Co-committed-by: Earl Warren <contact@earl-warren.org>
This is a followup of https://code.forgejo.org/forgejo/act/pulls/170 so that it is possible to read a workflow without validation. It is not uncommon for Forgejo to read a workflow just to extract a few information from it, knowing it has been validated before. It would be a performance regression if schema validation happened in these cases.
This is a port of https://github.com/nektos/act/pull/2717/files
It is a breaking change in the context of Forgejo and Forgejo runner because it will need to add the new `validate` argument when reading workflows.
Co-authored-by: ChristopherHX <christopher.homberger@web.de>
Reviewed-on: https://code.forgejo.org/forgejo/act/pulls/180
Reviewed-by: Michael Kriese <michael.kriese@gmx.de>
Co-authored-by: Earl Warren <contact@earl-warren.org>
Co-committed-by: Earl Warren <contact@earl-warren.org>
- do not override username and password when looping over services
- split prepareJobContainer out of startJobContainer
- split getNetworkName out as it is used by both
- add unit tests for prepareJobContainer
- make containre.NewContainer mockable
- add MockVariable helper
Closesforgejo/runner#575
---
Note to reviewers: do not show whitespace change, the refactor will show in a minimal way. When the fix is reverted the tests fail as follows:
```
Diff:
--- Expected
+++ Actual
@@ -81,4 +81,4 @@
Image: (string) (len=10) "some:image",
- Username: (string) (len=17) "containerusername",
- Password: (string) (len=17) "containerpassword",
+ Username: (string) (len=16) "service2username",
+ Password: (string) (len=16) "service2password",
Entrypoint: ([]string) (len=3) {
Test: TestStartJobContainer/Overlapping
```
Reviewed-on: https://code.forgejo.org/forgejo/act/pulls/181
Reviewed-by: Michael Kriese <michael.kriese@gmx.de>
Co-authored-by: Earl Warren <contact@earl-warren.org>
Co-committed-by: Earl Warren <contact@earl-warren.org>
- rc.getToolCache(ctx) is used to figure out RUNNER_TOOL_CACHE and returns RUNNER_TOOL_CACHE if it is found in the runner config, e.g.
```yaml
container:
env:
RUNNER_TOOL_CACHE: /srv/toolcache
```
- store the value in the new `toolCache` data member for containers, in the same way it is done for host
- GetRunnerContext for containers return `toolCache` instead of a hard coded string
- add integration test
Closesforgejo/runner#184
Reviewed-on: https://code.forgejo.org/forgejo/act/pulls/178
Reviewed-by: Michael Kriese <michael.kriese@gmx.de>
Co-authored-by: Earl Warren <contact@earl-warren.org>
Co-committed-by: Earl Warren <contact@earl-warren.org>
It is a breaking change because it changes how the shell is
determined.
Before, if `jobs.<job_id>.container.image` is set and the shell is not
specified, it defaults to `sh` instead of `bash`.
After, regardless of `jobs.<job_id>.container.image`, if the shell is
not specified, it defaults to `bash` if available, otherwise it
defaults to `sh`.
Rework the shell integration tests:
- Remove container specific tests because the special behavior related
to shell being set differently when a container image is present is
removed
- Modify the defaults test case to verify the fallback logic
- Use container images from code.forgejo.org to escape rate limiting
in the CI
- Add the missing node test
- Use
container:
image: code.forgejo.org/oci/node:22-bookworm
instead of
container: code.forgejo.org/oci/node:22-bookworm
because it silently failed to run (with no exit code)
- Prefer `-z "${BASH}"` because `-z ${BASH+x}` reads obscure
Closesforgejo/runner#150
- log job result as info not as debug
- add test
---
v6.4.0 regression introduced in 4880b091a2
It did not fail a test because the [original fix](https://code.forgejo.org/forgejo/act/pulls/67/files) has tests only for the case where a step is skipped, not when a job is skipped.
Closesforgejo/runner#660
Reviewed-on: https://code.forgejo.org/forgejo/act/pulls/172
Reviewed-by: Michael Kriese <michael.kriese@gmx.de>
Co-authored-by: Earl Warren <contact@earl-warren.org>
Co-committed-by: Earl Warren <contact@earl-warren.org>
https://github.com/nektos/act/pull/2372
+ fix: make node tool non volatile
Currently downgrading node via setup-node can break later actions
+ fix it and lookup on startup
+ fix problems
Co-authored-by: ChristopherHX <christopher.homberger@web.de>
Reviewed-on: https://code.forgejo.org/forgejo/act/pulls/168
Reviewed-by: earl-warren <earl-warren@noreply.code.forgejo.org>
Co-authored-by: Andrii Chyrva <andrii.s.chyrva@hotmail.com>
Co-committed-by: Andrii Chyrva <andrii.s.chyrva@hotmail.com>
Currently the only way to get pre and post actions is to go through the nodejs mechanism, which is quite wasteful when all one wants to do is run a couple of shell commands, I'm trying to get around this with this patch.
It works similar to the node* actions in that it supports `pre`, `main` and `post`.
It is different in that these strings are passed to the system shell using `sh -c` and execute similar to the composite run action with the shell set to `sh`.
Example action to make use of this patch: https://codeberg.org/slatian/test-action/src/branch/main/action.yaml
Reviewed-on: https://code.forgejo.org/forgejo/act/pulls/141
Reviewed-by: earl-warren <earl-warren@noreply.code.forgejo.org>
Co-authored-by: Slatian <baschdel@disroot.org>
Co-committed-by: Slatian <baschdel@disroot.org>
- upgrade to golangci-lint@v1.62.2
- make it renovate friendly
- remove most frequent lint check that are not of consequence (unused
args, etc.)
- fix remaining lint errors
- add renovate custom manager to update the Makefile variable
In the context of Forgejo the act-toolcache must not be mounted
otherwise independent workflows will race against each other when
writing to it. For instance if a setup-go action is run at the same
time as another, there is a probability that they both write the same
file at the same time, which could lead to a truncated or invalid
content.
The comment that introduced this change suggests it was motivated by a
border case by which the image would be empty. It is however unclear
why it should have any impact on how the network name is determined.
The hunk is reverted.
https://github.com/nektos/act/pull/1949/files#r1315163582
Implements one part of forgejo/runner#119. The other part is a corresponding PR in forgejo/runner: forgejo/runner#120.
Reviewed-on: https://code.forgejo.org/forgejo/act/pulls/24
Reviewed-by: earl-warren <earl-warren@noreply.code.forgejo.org>
Co-authored-by: s3lph <codeberg@s3lph.me>
Co-committed-by: s3lph <codeberg@s3lph.me>
Since the start script may create LXC templates that are shared, they
may race against each other when running for the first time. A lock
global to the host needs to be used to guarantee that does not happen.
