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

Re: [XP] Re: [Fit-users] Blog entry: A Vision for Fit

Expand Messages
  • yahoogroups@jhrothjr.com
    From: Brian Marick To: fit-users@lists.sourceforge.net
    Message 1 of 1 , Nov 2, 2005
      From: "Brian Marick" <marick.at.testing.com@...>
      To: "fit-users@..."
      Cc: "fit-devl@..."
      Sent: Wednesday, November 02, 2005 12:18 PM
      Subject: [XP] Re: [Fit-users] Blog entry: A Vision for Fit

      >I agree with this vision. What's needed to realize it?

      Some of these things are relatively easy to accomplish,
      some are already there (but not in batch Java).

      > - We don't have a common understanding of the human side of the
      > communication beyond things like "use a whiteboard" and "testers should
      > ask clarifying questions but take care not to overwhelm the
      > conversation". There's a lot more to learn, boil down, and share.
      > - The Customers are enthusiastic about the documents they get. The
      > table-writers (never, in my experience, the actual Customer person) are
      > less enthusiastic. There isn't a tool specialized for table-writing.


      However, that brings up an interesting question. How frequently
      do we have the actual customers writing tests? If the answer is
      effectively never (over the entire customer base, not just one
      consultant's customers) then the "feature" of allowing the customers
      to write in their own prefered tools seems to be excess baggage.
      I'd _really_ like a read on this.

      > Word's HTML is a horror, but that doesn't really matter. I don't find
      > it makes my table editing easy enough. Word for OS X (my version)
      > frequently messes up the formatting on tables (dropping grid lines,
      > most often) and once added an extra column.
      > - Programmers are the least enthusiastic. Here are the things I noticed
      > annoying programmers as I was showing them Fit on last week's
      > consulting gig:
      > + The packaging is not one-click, at least not if you want
      > FitLibrary.
      > (In my case, made worse by the fact that the latest FitLibrary
      > didn't
      > seem to work with the latest Java reference implementation. That
      > might have been my mess-up, but I'm pretty sure that problem
      > lost me one programmer before I reverted to my tried-and-true
      > older versions.)

      That ought to be relatively easy to fix, and the latest Fit for FitNesse
      does indeed not work with the batch FIT. It isn't supposed to.

      The Python install includes FitLibrary. I'm not sure if the current
      dotNet implementations or the Smalltalk implementation are going
      the integrated route or the separate package route though.

      I don't think we're going to see an integrated release since
      Rick wants to be able to release to his own schedule, and
      I don't blame him one bit. If I was in his position, I'd want
      that too.

      > + IDE integration is weak compared to JUnit. You can run the test in
      > one click, but it takes more than one click to find out if the
      > row you
      > expected to go green did go green. With JUnit, it's easy to point
      > at a test or test method, point at it, tell the IDE to run just
      > that one,
      > and thereafter run it again with one click. The equivalent for
      > Fit would
      > be to do that for the current page you're working on.

      IDE integration is basically non-existant, except for Hugo's efforts
      on Fitster/Eclipse.

      > + The worst thing seems to be editing the tables. It seems that we
      > frequently had cause to update the tables to capture ideas
      > discovered
      > in coding. (Example: calling across the room to ask the Customer
      > how many digits of precision to display for X.) When programmers
      > compare editing a Fit test to editing a JUnit test, well...

      I think we need a bit of requirements analysis here. See a more
      extended comment below.

      > + Lack of templates. I watched someone create a JUnit test with five
      > keystrokes in Eclipse ('t' - 'e' - 's' - 't' -
      > control-shift-meta-cokebottle).
      > There's no such aid for common tasks.

      The basic point here is that except for FitNesse, which is not
      integrated in any developer's IDE, and Fitster/Eclipse, there
      isn't any support for editing FIT tests in an IDE, popular
      or not.

      JUnit integration into Eclipse is very easy in comparison,
      because you can leverage off of the existing Java editor.
      I'm not a Java person and I don't know Eclipse well enough
      to know if there's a suitable HTML editor to start with.

      The IDE "market" is quite fragmented. There isn't an
      editor for Python that's really a front-runner, unless
      you count emacs, and I'd rather not. Effort integrating
      into a particular editor is essentially effort wasted since
      it won't be useful to more than 10% of the community,
      if that.

      John Roth
      Python FIT.
    Your message has been successfully submitted and would be delivered to recipients shortly.