Maybe 3rd time is the charm!
This makes the CI build run on windows too, and fix broken tests/featuers on Windows.
Most of the changes are related to ignoring certain tests on windows, or making sure to use the right path separators.
More work to do in the future, probably!
#4293
---------
Signed-off-by: Carlos Alexandro Becker <caarlos0@users.noreply.github.com>
<!--
Hi, thanks for contributing!
Please make sure you read our CONTRIBUTING guide.
Also, add tests and the respective documentation changes as well.
-->
<!-- If applied, this commit will... -->
...
<!-- Why is this change being made? -->
...
<!-- # Provide links to any relevant tickets, URLs or other resources
-->
...
---------
Signed-off-by: Carlos Alexandro Becker <caarlos0@users.noreply.github.com>
The PR enables
[`nolintlint`](https://golangci-lint.run/usage/linters/#nolintlint)
linter and fixes up appeared issues.
Changes:
- Enable `nolintlint` in `.golangci.yml` config.
- Remove unused `//nolint:` comments.
- Fix `nolint` comment format by removing spaces (`// nolint: dupl` ->
`//nolint:dupl`)
Co-authored-by: Carlos Alexandro Becker <caarlos0@users.noreply.github.com>
The PR cleans up unnecessary loop variable copying and enables the
[`copyloopvar`](https://golangci-lint.run/usage/linters/#copyloopvar)
linter for detecting this redundant variable copying.
#### Additional notes
After the project upgraded to Go version 1.22 in #4779, copying
variables inside a `for` loop became unnecessary. See this [blog
post](https://go.dev/blog/loopvar-preview) for a detailed explanation.
The `copyloopvar` linter is only available from `golangci-lint` v1.57
onwards, so we also need to update this tool.
Currently, GoReleaser will assume you're running against github, gitea
or gitlab.
You could set `release.disable: true`, but it would still set and try to
use some defaults that could break things up.
Now, if you disable the release, goreleaser will not set these defaults.
It'll also hard error in some cases in which it would happily produce
invalid resources before, namely, if `release.disable` is set, and, for
example, `brews.url_template` is empty (in which case it would try to
use the one from the release, usually github).
closes#4208
The check to skip Scoop releases when SCM ones are disabled seems to be
the only place where this is happening and in conflict to Brew releases.
These two need not be linked as users can configure the `url_template`
to point elsewhere.
---
Consider the following use case:
```
* goreleaser/my-app.yml
- builds: spec for building a Go binary for my-app
- dockers: spec for packaging my-app in a container and pushing to Artifactory & GCP
- archives: spec for zipping up my-app binary
- artifactories: spec for pushing archives of my-app to a private jfrog artifactory
- brews: spec for publishing my-app through brew on macOS with URLs from artifactory
- scoop: spec for publishing my-app through scoop on Windows with URLs from artifactory
- release: disabled SCM (e.g. GitHub) release
```
Running `goreleaser release -f goreleaser/my-app.yml` with the above
setup results in the correct artifacts (archives & containers) being
published to Artifactory, GCP, Brew, but no Scoop.
Given that Scoop's `skip_upload` cannot be templated in 1.18 and I do
not wish to have SCM releases & tags on every release, one can be rather
stuck.
mimic the structure to be created in the target repo, and put each of
them in its own folder.
Signed-off-by: Carlos Alexandro Becker <caarlos0@users.noreply.github.com>
closes#3989
Basically, when some of these pipes fail, the error will be memorized,
and all errors will be thrown in the end.
Meaning: the exit code will still be 1, but it'll not have stopped in
the first error.
Thinking of maybe adding a `--fail-fast` flag to disable this behavior
as well 🤔
---------
Signed-off-by: Carlos Alexandro Becker <caarlos0@users.noreply.github.com>
It should make goreleaser easier to use, as a `repository` is now always
called `repository`.
Signed-off-by: Carlos Alexandro Becker <caarlos0@users.noreply.github.com>
This allows to open pull requests across repositories on nix, brew, krew
and scoop.
closes#4048
---------
Signed-off-by: Carlos Alexandro Becker <caarlos0@users.noreply.github.com>
- log keys will be ordered as intended instead of sorted
- paths always relative to cwd
---------
Signed-off-by: Carlos Alexandro Becker <caarlos0@users.noreply.github.com>
This brings the scoops feature a bit more closer to similar pipes, like
brew and krew.
- It now supports multiple scoops
- It improves some validations to prevent wrong manifests
- It uses extra.binaries instead of extra.builds, as brew does too
- extra.builds is now unused, will be removed in a subsequent PR
- More tests were added as well
closes#3941
---------
Signed-off-by: Carlos Alexandro Becker <caarlos0@users.noreply.github.com>
basically redoing #3559 as it got impossible to merge with the many
changes since it was open (which is totally my fault for not merging it
earlier).
Anyhow, still a WIP, going also doing some other related improvements in
the way.
cc/ @graytonio
closes#3559closes#3525
---------
Signed-off-by: Carlos Alexandro Becker <caarlos0@users.noreply.github.com>
Co-authored-by: Grayton Ward <graytonio.ward@gmail.com>
alternative to #3806
the idea is that both `context.New` and `context.Context{}` are never
used in tests.
not sure yet how much I like it, so far code does look a bit more
readable though.
---------
Signed-off-by: Carlos A Becker <caarlos0@users.noreply.github.com>
- added a tmpl.Bool that checks if the result of a template is "true"
- added an ErrSkipper interface so Skip() methods can return errors as well
---------
Signed-off-by: Carlos A Becker <caarlos0@users.noreply.github.com>
The replacements thing was always a bit weird, especially on archives.
We can solve that with templates, so, removing I'm deprecating it.
Also did the same on other places that had it the same feature.
Closes#3588
Signed-off-by: Carlos A Becker <caarlos0@users.noreply.github.com>
closes#3508
- adds template support for krew and scoop repo refs
- template branch on reporef on brew as well
Signed-off-by: Carlos A Becker <caarlos0@users.noreply.github.com>
It was just allowing v2 and v3 due to some misreading on my side.
This commit fixes it to allow v1, v2, v3 and v4.
refs #3016
Signed-off-by: Carlos A Becker <caarlos0@gmail.com>
* feat: allow to template commit author
Signed-off-by: Carlos A Becker <caarlos0@gmail.com>
* chore: fmt
Signed-off-by: Carlos A Becker <caarlos0@gmail.com>
