Spotify Labs Blog on Testing Microservices

(Simon Godfrey) #1

So I just started reading this blog and straight away I was hit by this statement.

Anyone who has ever done manual testing knows that it’s slow, boring and error-prone.

I believe they are talking about scripted testing but that is not “manual testing”. I’m horrified by that statement.

(João Farias) #2

“Most people would say that we write tests to verify that things work as we expect them to… After all, that can be verified through manual tests as well.”

Apparently, he used “test” as synonym for James Bach’s “check”, something that you know and want to verify - not Bach’s “test” definition, a process of learning and exploring.

Given the blog’s definition of testing, I would say that

“Anyone who has ever done manual testing knows that it’s slow, boring and error-prone.”

is pretty accurate and should be done by machines as much as possible.

(Kate) #3

I don’t know about everyone else, but I design tests to

  • Verify that things work as I expect them to
  • Verify that the system handles unexpected things with a degree of grace
  • Verify that users aren’t going to be asked to do something really confusing/weird/counter-intuitive
  • See if I can break the system in a way that exposes security flaws.
  • See if I can break the system in a way that a user is likely to act.
  • Verify that other important features are still working as designed.

That’s not a complete list, but it covers most of what I do - and only the last item would count as “slow, boring, and error-prone” when done manually. For a new development, exploring the changes may be slow, but it’s certainly not boring, and the error-proneness usually has more to do with factors that have nothing to do with me.

I got to that point of the blog, and immediately dismissed the article as “let’s sell the rubes a tool that might or might not do what they need”.

(Maggie) #4

What gets me is how these kinds of blogs promote misconceptions of what testing is. It makes it sound like as if the work that I and many other people do is slow, boring, and outdated. I mean, automation is a good tool for regression testing but that’s not the only kind of testing that has to be done. Imagine if I were to write a blog saying “Hey, automation exists solely to devalue testing and make it easier for companies to hire less testers .” It would be dismissed as ridiculous.