CODEWITHSUNDEEP
testinggo
Varun Patro
Varun Patro

Reputation: 2179

How to run golang tests sequentially?

When I run go test, my output:

--- FAIL: TestGETSearchSuccess (0.00s)
        Location:       drivers_api_test.go:283
        Error:          Not equal: 200 (expected)
                                != 204 (actual)

--- FAIL: TestGETCOSearchSuccess (0.00s)
        Location:       drivers_api_test.go:391
        Error:          Not equal: 200 (expected)
                                != 204 (actual)

But after I run go test again, all my tests pass.

Tests fail only when I reset my mysql database, and then run go test for the first time.

For every GET request, I do a POST request before to ensure that there is data created in the DB.

Could anyone help me with how to make sure that tests are run sequentially? That is the POST requests are run before the GET requests?

Upvotes: 40

Views: 68717

Answers (5)

baal_imago
baal_imago

Reputation: 313

It's also possible to synchronize the test using wait groups:

    awaitRootElement := sync.WaitGroup{}
    awaitRootElement.Add(1)

    t.Run("it should create root element", func(t0 *testing.T) {
        // do the test which creates root element
        awaitRootElement.Done()
    })
    awaitRootElement.Wait()
    t.Run("it should act on root element somehow", func(t0 *testing.T) {
      // do tests which depend on root element
    })

Note that you should wait before scheduling the tests, since the asynchronous test execution otherwise might deadlock (the test routine is awaiting another test which never gets to run).

Upvotes: 3

Alik Khilazhev
Alik Khilazhev

Reputation: 1107

For those who as I am is getting problems because of multiple concurring tests running simultaneously. I found a way to limit the maximum number of test running in parallel:

go test -p 1

With this, your test will run sequentially one by one.

Source

Upvotes: 47

icza
icza

Reputation: 417412

You can't / shouldn't rely on test execution order. The order in which tests are executed is not defined, and with the use of testing flags it is possible to exclude tests from running, so you have no guarantee that they will run at all.

For example the following command will only run tests whose name contains a 'W' letter:

go test -run W

Also note that if some test functions mark themselves eligible for parallel execution using the T.Parallel() method, the go tool will reorder the tests to first run non-parallel tests, and then run parallel tests in parallel under certain circumstances (controlled by test flags like -p). You can see examples of this in this answer: Are tests executed in parallel in Go or one by one?

Tests should be independent from each other. If a test function has prerequisites, that cannot be done/implemented in another test function.

Options to do additional tasks before a test function is run:

  • You may put it in the test function itself
  • You may put it in a package init() function, in the _test.go file itself. This will run once before execution of test functions begins.
  • You may choose to implement a TestMain() function which will be called first and in which you may do additional setup before you call M.Run() to trigger the execution of test functions.
  • You may mix the above options.

In your case in package init() or TestMain() you should check if your DB is initialized (there are test records inserted), and if not, insert the test records.

Note that starting with Go 1.7, you may use subtests in which you define execution order of subtests. For details see blog post: Using Subtests and Sub-benchmarks, and the package doc of the testing package.

Upvotes: 47

CESCO
CESCO

Reputation: 7758

Apart for 3rd party libraries like Convey and Ginkgo, with plain Golang 1.7 you can run tests sequentially. You can read more here

func TestFoo(t *testing.T) {
    // <setup code>
    t.Run("A=1", func(t *testing.T) { ... })
    t.Run("A=2", func(t *testing.T) { ... })
    t.Run("B=1", func(t *testing.T) { ... })
    // <tear-down code>
}

And you can run them conditionally with:

go test -run ''      # Run all tests.
go test -run Foo     # Run top-level tests matching "Foo", such as "TestFooBar".
go test -run Foo/A=  # For top-level tests matching "Foo", run subtests matching "A=".
go test -run /A=1    # For all top-level tests, run subtests matching "A=1".

So lets say you got an user package from a REST api that you want to test. You need to test the create handler in order to be able to test the login handler. Usually I would have this on the user_test.go

type UserTests struct { Test *testing.T}
func TestRunner(t *testing.T) {

    t.Run("A=create", func(t *testing.T) {
        test:= UserTests{Test: t}
        test.TestCreateRegularUser()
        test.TestCreateConfirmedUser()
        test.TestCreateMasterUser()
        test.TestCreateUserTwice()
    })
    t.Run("A=login", func(t *testing.T) {
        test:= UserTests{Test: t}
        test.TestLoginRegularUser()
        test.TestLoginConfirmedUser()
        test.TestLoginMasterUser()
    })

}

Then I can append methods to the UserTest type that wont be executed by the go test command in any _test.go file

func (t *UserTests) TestCreateRegularUser() {
    registerRegularUser := util.TableTest{
        Method:      "POST",
        Path:        "/iot/users",
        Status:      http.StatusOK,
        Name:        "registerRegularUser",
        Description: "register Regular User has to return 200",
        Body: SerializeUser(RegularUser),
    }
    response := util.SpinSingleTableTests(t.Test, registerRegularUser)
    util.LogIfVerbose(color.BgCyan, "IOT/USERS/TEST", response)
}

Upvotes: 26

nouney
nouney

Reputation: 4411

The best way to achieve that is to create a TestMain, as presented here. 

import (
  "testing"
  "os"
)

func TestMain(m *testing.M) {
   // Do your stuff here
   os.Exit(m.Run())
}

Upvotes: 3

Related Questions