permaculture playing cards
The moose likes Testing and the fly likes how to test drive this problem? Big Moose Saloon
  Search | Java FAQ | Recent Topics | Flagged Topics | Hot Topics | Zero Replies
Register / Login
JavaRanch » Java Forums » Engineering » Testing
Bookmark "how to test drive this problem?" Watch "how to test drive this problem?" New topic

how to test drive this problem?

Piotr Trochim
Ranch Hand

Joined: Dec 29, 2006
Posts: 35

I have the following story to implement - I need to be able to build a scene in a game based on its predefined representation.

On one hand I have my scene - that contains the definition of:
- the terrain
- lights
- cameras
- players

On the other hand, there's this factory that knows just how to "create" them in the 3D engine I'm using.

Now - I tried several approaches, and each time I finish the refactionring after the first iteration, I end up with stuff I started

Would somebody please be willing to show me how to evolve the whole thing through the tests?
If so, then start evolving the code from scratch - meaning that you should evolve the scene definition and the factory I mentioned as well.

Thank you so very much

Jeanne Boyarsky
author & internet detective

Joined: May 26, 2003
Posts: 33130

Why don't we start with what you did. What kinds of tests did you write?

[OCA 8 book] [Blog] [JavaRanch FAQ] [How To Ask Questions The Smart Way] [Book Promos]
Other Certs: SCEA Part 1, Part 2 & 3, Core Spring 3, TOGAF part 1 and part 2
Piotr Trochim
Ranch Hand

Joined: Dec 29, 2006
Posts: 35

Here is one of the approaches I had - it's not that bad and I think it is a way to go, but if you could take a look at it and tell me what did I screw up, that'd be great.

A word of information - in the Red Phase I always use pseudocode to get my idea down onto the paper, then I translate it and make it work during the Green Phase.

Iteration 1, phase Red

I want to be able to create something on my scene - let that something be a terrain.

... hmmmm - how to test that... - I don't want to get the implementation of how I create a terain in a particualr 3d Engine involved yet...

... Yet another approach ....

.. ok - at least this time I had a clear testing point...

Iteration 1, phase Green

The shortest steps to get things working....


Iteration 1, phase Refactoring

I messed it up quite badly - the code smells.

The Terrain class is just a marker - it's used only in the test - I'll remove it for now

Ok - seems kinda clean. I have my scene definition captured in one class, the way it should be built in the other ... Let's move on.

Iteration 2, phase Red

I want to be able to add not only the terrain, but also the light to the scene.

Iteration 2, phase Green

Ok - it's working. Now for some refactoring...

Iteration 2, phase Refactoring

I don't like it - Open-Close principle's good name is being dragged throught the mud here.
I''m expecting all kinds of different entities here, and I can't keep adding methods each time a new one comes up...

One more thing - I noticed that I needed to introduce exactly the same change to the factory as the one I introduced to the SceneDefinition.
Maybe their responsibilities are not so well separated after all...

Basically, after applying DIP I get the following code, where instead of having a factory, I end up with a bunch of creators that can handle particular objects creation process.

Everything works, and now I don't have to worry about breaking the OCP any more.

BUT - what I ended up with - I mean the SceneDefinition - that's just a wrapping for a list - does nothing more but store the creators and run a create on them all - it resembles more of a Compound then some separate entity - so I'll evolve it towards that...

Ok - now that's more like it

So basically that's my story. I can now create those things dinamically based on some arbitrary representation using parsers etc..

What do you think? Is it ok?

I agree. Here's the link:
subject: how to test drive this problem?
It's not a secret anymore!