Re: [Junit-devel] ensuring that tearDown code gets run

[Dossy <do...@pa...>]

On 2002.05.31, Kent Beck <ke...@th...> wrote:
> Every time I've wished for shared setup because of performance and I've
> looked to refine and further decouple the design instead, I've been glad I
> did. Of course, this wouldn't work for you.

Do you ever use JUnit to do acceptance testing of "the real thing"?

In particular, I'm working with a vendor who's building an interface
to their application's data via a TCP server and XML.  I began coding
a whole suite of Acceptance Tests of their implementation of the
server (after writing similar tests for my client, where I mocked
out the server with an in-memory spoof).  However, in order to test
the server, I _have_ to use my client, go out over the network,
speak to their server and interpret their response against my
canned (expected) responses.  I cannot test their code in-memory
since I have no access to their code, just their exposed interface.

I found JUnit to be a big pain, so I ended up moving everything
to Ruby and doing my testing with Test::Unit in Ruby.  I'm much
happier.

Maybe the answer here is two-fold:  1)  When testing needs shared
setup, make sure it really actually needs it.  Try to design your
tests so that it doesn't -- even if it means more work today, it
really pays off if you can avoid it.  However, 2)  When testing
actually needs shared setup, use a testing tool other than JUnit.
It's just not the right tool for the job, so learn to recognize
that rather than how you might otherwise hack kludges into JUnit.

JUnit does what it does and it does it well.  However, it doesn't
do very much else without customizations.  There's absolutely
nothing wrong with this: there are dozens of testing frameworks
available -- there has got to be one that's a good fit for your
problem.  Have multiple tools in your toolbox so that everything
does NOT end up looking like a nail ...

-- Dossy

-- 
Dossy Shiobara                       mail: do...@pa... 
Panoptic Computer Network             web: http://www.panoptic.com/ 
  "He realized the fastest way to change is to laugh at your own
    folly -- then you can let go and quickly move on." (p. 70)