Every body seems to be talking about TDD and BDD these days so i thought i give it a try on a smallish home project.
A short backgrund I am developing a class Device in a C extension that interacts with a native C API for controlling remote devices like light switches and so on.
To make the tests more platform independent I have mocked out the C API to behave as the original one but instead of controlling hardware it sets a global ruby state variable (a hash containing all the devices).
This is all fine and works as expected. I can compile and link to this library instead of the real one.
Now the part i need help with is how to test this following some sort of best practice I have the following spec to describe my Device class
describe Device do
describe ".get" do
context "Specified id exists" do
it "Returns a instance of Device"
it "Device has same id as specified"
end
context "Specified id is undefined"
it "throws UndefinedDeviceException"
end
end
describe "#turn_on" do
context "Device supports turning on" do
# I set up the mock state to have a device that supports on..
# how do i get the Device instance here? is it ok to depend on
# Device.get method?
it "Returns true"
it "Set device state to on" # Check mocked state
end
end
.....
end
Now as far as i understand in each context i setup the state for my mocked library to reflect what I want to test and i invoke the method.
This works well for the static method Device.get(id) because this returns a new Device what I'm not sure on how to test is the instance methods. If I want to test the method turn_on i need to have a instance of Device first which means i need to call Device.get to actually get the device? but if that's the case arnt I again "testing" the class method get? I guess what I'm really asking here is whats a good practice to get a instance of a object in Rspec if it depends on an external library.
