On 19 February 2010 08:59, Erik P. [email protected] wrote:
creating more debugging time for myself than I’m saving without
I think you can consider writing a completely separate spike to explore
bar etc. or perhaps you need to go more granular and create some tools to
simplify your calculations. Finally if you have any long methods in your
calculations (> than 5 lines) refactor them. Decomposing them into smaller
methods could reveal alot about the structure of your problemHTH
Andrew
I’m a bit with Andrew on this. I think you are not asking the right
question.
In TDD/BDD I view writing the tests/spec as PART of writing the code, so
if
you are questioning the value of spending the time to write the specs,
and
lamenting that you feel this is wasted effort, I have to ask why are you
not
questioning the value of even writing code when the target is changing
so
frequently? Why don’t you feel that the code itself that you’ve had to
discard/rewrite/refactor so much has not also been wasted effort? How
is
what you are doing going to deliver value to the end user?
If this is a case where someone else can’t make up their minds what they
want, then potentially it might be a situation where writing the code
itself
is just a waste of time, much less writing the specs, and you need to
spend
your effort elsewhere until what is needed is better known.
OTOH, if this is a situation where you are doing a lot of
experimentation
and discovery, and the writing of the code is necessary to figure out
‘what
will work’, then the writing of tests might not add value and just being
slowing you down. I’d base that judgement on the degree to which the
code
itself is considered disposable and the likelyhood that it won’t ever
make
it into production ‘as is’. In that case, then there may not be much
value
to fleshing out anything but the high level spec definitions (just as a
means of keeping straight what you are trying to do this iteration, and
potentially a type of documentation to others of what ‘approach number
27’
is supposed to take for inputs and produce for outputs.
Kent Beck touches on this in his ‘flight of the startup’ blog post
http://www.threeriversinstitute.org/blog/?p=251 and a followup
‘developing
for the flight of the startup’
http://www.threeriversinstitute.org/blog/?p=252 and he makes some good
points about when it pays to be dogmatic about some agile practices such
as
TDD, and when it’s better to put some of those practices aside depending
on
what it is that you are trying to deliver to the customer at a
particular
point in the project.
If you are currently in the phases Kent describes as Taxi and Takeoff,
then
you could very well be right to dispense with a majority of tests and
other
practices that are needed for maintainable high quality code. (because
you’re not trying to deliver highly maintainable code.) OTOH if you are
in
the Climb or Cruise stages, then I’d be greatly concerned that ‘what we
are
building’ is in such a constant state of flux, and I’d suggest something
has
a smell that’s ungood.