Fail within ginko testing not os.Exit(1)

[cdrage]

We should be failing using: https://onsi.github.io/ginkgo/#marking-specs-as-failed within our end-to-end tests rather than using: https://github.com/redhat-developer/ocdev/blob/master/tests/e2e/e2e_test.go#L41

I suggest that we modify https://github.com/redhat-developer/ocdev/blob/master/tests/e2e/e2e_test.go#L28 to return err and if it errors, use Fail("Failure reason") with the stderr output from the ocdev command.

[cdrage]

ping @ashetty1

[cdrage]

Just had a discussion with @ashetty1 on bluejeans, So after looking at this issue: onsi/gomega#115

I think we should consider using / trying out this code:

command := exec.Command(pathToLs, "")
session, err := gexec.Start(command, GinkgoWriter, GinkgoWriter)
Ω(err).ShouldNot(HaveOccurred())

commandReal := exec.Command("ls")
sessionReal, err := gexec.Start(commandReal, GinkgoWriter, GinkgoWriter)
Ω(err).ShouldNot(HaveOccurred())

session.Wait() // waits for the session to exit
Eventually(sessionReal).Should(gexec.Exit(0)) // equivalent to `session.Wait()` but allows you to assert on exit code

Ω(session.Out.Contents()).Should(Equal(sessionReal.Out.Contents()))

and see if it helps. Seems to fit our use-case exactly and makes things a bit more simpler instead of using runCmd()

[kadel]

When this issue gets fixed, please remove the verbose flag that was added in #308.

[cdrage]

Closing this as #320 was merged and then #322 (to remove the verbose output).

Closing this issue!