Ron,

--- In extremeprogramming@yahoogroups.com, Ron Jeffries
<ronjeffries@...> wrote:
> Distinction without a difference? Or would you care to say what
> problem they are symptoms of? Anyway, doesn't matter to my point,
> which is that acceptance tests can help address those issues.

There is a difference and I am asking what you (and others) think the
root causes are that are somehow addressed by acceptance tests.

I suppose I could accept that bugs are the real problem and there is no
need to look any deeper.

I have a feeling if I ever suggested on this list that there was no need
to search for root causes, you and others would loudly object. :)

> I think the OP was asking you how you manage to avoid those issues
> without acceptance tests. How do you?

I asked first! :) In all honesty, I am not being Socratic... I am
trying to figure this out myself.


> Anyway, if you're not having those issues listed, we're wondering
> what you think is helping you to avoid them ... besides just being
> incredibly smart and careful. And if you are ... well, you know what
> we might recommend. But I'm guessing you aren't, and wonder why. :)

Well if I was incredibly smart I wouldn't be asking, I would be telling!

We had a discussion this afternoon on our team and with the boss. We
discussed the possibility that Rob mentioned... that having a single
lead developer across the lifespan of the project can help avoid
pitfalls. This might be true... and if so, it's not very Agile.

The other possibility is that the acceptance tests that we *do* have are
around the high risk, core business functionality. But even those have
proven marginal in value.

Another topic that came up was small team size. The development team
has ranged from 2 to at most 5 developers. This eliminates a lot of
communication issues although I am not convinced that communication is a
root cause of many bugs.

Interestingly, we all feel that Fitnesse and Watin tests both lack the
true "end to end" testing that would give us the level of comfort that
many people get from acceptance tests. We aren't against acceptance
tests... we just want more for our money than we are currently getting!

Ultimately, what I hope will come from this conversation is an
understanding of what root problems are addressed with acceptance
testing. If we understand that better, we can tailor our solutions to
the specific problem(s) that we are most prone to. We can also justify
throwing money at a problem if we can identify it.

Thanks,

Matt