Loading ...
Sorry, an error occurred while loading the content.

108730Re: [XP] Re: Running Tested Features

Expand Messages
  • Ron Jeffries
    Jul 15 3:47 AM
      On Friday, July 15, 2005, at 3:36:20 AM, PierG wrote:

      > --- In extremeprogramming@yahoogroups.com, Ron Jeffries
      > <ronjeffries@X...> wrote:

      >> > . how do you 'specify' the acceptance test?
      >> Might write a GUI-driving test.
      >> Might just test it manually.
      >> Might test the "SaveAs" method with an automated test.

      > Might be manual tested???
      > Well ... one of the key factors, in my opinion, it that it's
      > automatic. Am I wrong?

      Yes, no. You asked how I might specify the test. I might just try
      it. I don't expect the GUI to go wrong once it is right, because my
      experience is that one I get them to work I don't break them. So my
      judgment, good or bad, might be that I would test it manually, then
      leave it alone.

      If I were really good at automating GUI tests, I might choose
      differently. But I'm not, and I have to choose where and how to
      invest my time in testing, coding, and learning. Right now, my
      priorities point elsewhere. If I could spawn a subprocess to work on
      this, I would.

      Is it troubling that even the most absolute rule -- only automated
      test count -- isn't absolute in my view?

      >> > . tools to use in .Net environment?
      >> NUnit. Fit/FitNesse.
      > Well .. NUnit is something usually used for unit testing/TDD or at
      > least is how we use it. Are Fit/Fitness supporting GUI-Driven tests?

      Not very well, the way I do them. I don't like GUI-driven tests, see
      above. I don't know of any way at all to do them that doesn't seem
      to me to be a waste of time.

      I keep hoping that someone will come up with a way that I can
      understand and use, but so far I've not been exposed to one that

      Furthermore, I hold the view that GUI-driven thinking is at the root
      of a lot of bad code, so I try really hard to think about the
      "model" and let the GUI be a view of the model. When I do that, I
      seem to get a better design, and it also means that I have much less
      need to do automated testing of the GUI.

      I took you to be asking what I'd do, and answered that. My answer
      does not have the form you expected ... or the form that I wish it
      could have, namely: here's this cool automated way to test Save As.

      We need some smarter person here. Meanwhile ... those really are the
      things I'd consider doing. Best I've got right now. :)

      Ron Jeffries
      You have to either laugh or cry. -- Bill Rogers
    • Show all 24 messages in this topic