Re: Another Little 2.5 Data Testing Issue

On Fri, May 9, 2008 at 5:14 PM, Charlie Poole
<[email protected]> wrote:
> Hi Kelly,
>
> TestFixtureSetUp and friends ( TFTD, SU, TD ) all have to do
> with running tests. The extensions have to do with recognizing
> and loading tests.

I understand the model, but I didn't understand the model very well
until recently. I suspect that this could be confusing to the average
user, and because of the increased complexity, there probably should
be at least a little more documentation aimed at how things work
overall, rather than just the "leaf-oriented" documentation that we
have now.

I'll happily write something up if there is a place to put such things.

> If you think about it, this I logical: you can't be running
> any kind of SetUp for a test until you actually have a test.

I know that it makes sense to you Charlie. One of the difficult things
after you've been associated with a project for a while is to feel
empathy for the newbie user. I'm still new enough that I remember
being clueless. :-)

> I'd like to also have extension points that are executed when
> the test is running, but we don't yet have them in this beta.

Ok... not exactly sure what you're getting at here, but it sounds interesting.

> There's a bit of an issue, however, as to how to distinguish
> the different cases - not for NUnit, which is easy, but for
> the user. We need some names that make sense.

Not just names, but also documentation, I think.

> At first, I was thinking that different data extensions would
> work differently. But now I lean toward allowing (for example)
> TestCase to be interpreted either at load time or at run time.
> We could distinguish the two by having a differet type of
> test - say a DynamicTest.

I'll have to think more about this later. Not computing entirely at the moment.

> Be sure to test what we have, however, in light of what it does.
> We want to discuss alternatives, but we also want to be sure
> that what's already published works as advertised.

I'm trying to make it work, and to try out different approaches. The
next thing I'm going to try is the class oriented approach.

My observation thus far is that DataSource is more flexible than
IterativeTest, but it also is more complicated to write an IEnumerable
for since you have to pack everything into an object[] for each call.
I understand the decision and tradeoff, but for the simplest case
(which is what people will do most of the time) I think the
IterativeTest is easier, but not as flexible. Don't get me wrong, I'm
really glad there is something in NUnit itself so I don't have to
maintain it myself forever :-) even if it does take some getting use
to.

Will continue to try things out over the next week.

-Kelly

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
Don't miss this year's exciting event. There's still time to save $100. 
Use priority code J8TL2D2. 
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone