Replace gopkg.in/check.v1 with the standard testing library

[carloslima]

closes #21

[Copilot]

Pull Request Overview

This PR removes the dependency on gopkg.in/check.v1 and converts all existing test suites to use Go’s standard testing package, preserving test logic and adding subtests or reflect.DeepEqual/regexp‐based assertions.

• Deleted the legacy suite_test.go and go-check suite boilerplate.
• Updated multiple *_test.go files to replace GoCheck assertions with testing.T methods and reflect.DeepEqual/regexp checks.
• Cleaned up go.mod by removing the GoCheck requirement.

Reviewed Changes

Copilot reviewed 6 out of 7 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
suite_test.go	Removed GoCheck suite setup (no longer needed).
node_test.go	Converted node tests to use testing.T, reflect.DeepEqual, regexp.
limit_test.go	Switched limit tests to testing.T and regexp, removed GoCheck.
encode_test.go	Migrated encoding tests to testing.T, added subtests.
decode_test.go	Migrated decoding tests to testing.T, added subtests.
go.mod	Removed gopkg.in/check.v1 from module requirements.

Comments suppressed due to low confidence (4)

encode_test.go:515

• The fmt.Sprintf call is used here but fmt is not imported; add import "fmt" to the import block.

		t.Run(fmt.Sprintf("test %d: %q", i, item.data), func(t *testing.T) {

decode_test.go:870

• The fmt.Sprintf call is used here but fmt is not imported; add import "fmt" to the import block.

		t.Run(fmt.Sprintf("test %d: %q", i, item.data), func(t *testing.T) {

decode_test.go:1954

• bytes.NewBuffer is used here but bytes is not imported; add import "bytes" to the import block.

		dec := yaml.NewDecoder(bytes.NewBuffer([]byte(item.data)))

limit_test.go:43

• [nitpick] Instead of return to skip short tests, call t.Skip("skipping long test in short mode") so the test is marked as skipped.

}

[ccoVeille]

I raised a few things, but LGTM 👍

Thanks for working on this

[andig]

That's why the whole thing felt so odd. I've also made good experience with github.com/stretchr/testify to complement testing. This pattern

	if err != nil {
		t.Fatalf("Close() returned error: %v", err)
	}

then becomes

    require.NoError(t, err, "Close() returned error: %v", err)

or simply

    require.NoError(t, err, "Close() returned error")

[ccoVeille]

@andig I'm unsure what you meant.

The way it sounds to me is "okay, but it was simpler with testify and I'm used to it."

While I agree, it doesn't sound logic to me.

Dolmen and @brackendawson testify maintainers asked yaml project to do not depends on testify because testify depends itself on (the old) yaml project. The idea is to avoid cyclic dependencies.

• Refactor tests to use vanilla testing package plus github.com/stretchr/testify #21 (comment)
• Refactor tests to use vanilla testing package plus github.com/stretchr/testify #21 (comment)
• Refactor tests to use vanilla testing package plus github.com/stretchr/testify #21 (comment)

So testify cannot be used.

Here I don't see a problem with this PR. A few files are updated, yes, it's not as simple as testify, yes, it is using vanilla testing.

But I would say it's expected because:

• it was what was asked the issue
• it is what this PR is about.

If there is a huge problem, nothing prevents to add a internal wrapper library that could check:

• for error
• if error is equal to another (using errors.Is)
• if error contains a string

And report test failure message returning the expected state.

It's easy. It can be extended if needed.

But this would be a NIH logic,
an/or at real choice of reducing the number of dependencies.

BTW Bracken suggested to use is which has no dependencies. It could be considered.

#21 (comment)

I feel like having a few wrappers could be considered, if vanilla become a problem.

This is the reply I'm doing about what I understood you say. But as I said, I might have misunderstood what you meant. So please apologize myself if I'm wrong and my reply is pointless.

[andig]

The idea is to avoid cyclic dependencies.

Sorry, I wasn‘t aware of that. I‘m confused by how that would be a bad thing for tests, but that‘s OT. No concerns, sorry for the noise.

[stefanprodan]

LGTM

PS. We should be using t.Setenv but this can come later in a different PR.

[ingydotnet]

This looks good enough to merge.

We can apply the optimization ideas in further PRs.