Saturday, September 7, 2019

Test Driven Deck

While having a plan for testing your code is important, Test Driven Development (TDD) takes this to another level. The idea here is that you write automated tests before you start writing the code. When it is practical to do this, it is an incredibly good way of creating code. Planning tests gets you thinking about the code you are going to write – hopefully with an idea of the fringe cases that you are going to encounter. Writing the tests give you something to check the code against. All tests should initially fail, with a successful test likely indicating either an error with the test or an error with the test plan.

You then write the code and run it against the tests trying to pass one test at a time. Once you have passed all the tests and no missing tests have been discovered, you are finished the code and can start planning the tests for the next piece of code that you need to write. Having automated tests also means that it is easy to have the tests running frequently while you do other things and if something breaks then you know right away and as it is likely what you (or someone on your team) is currently working on, it will be trivial to fix the bug! Remember that in general the quicker you can find a bug, the less it will cost to fix it. This cost is in time, but can also be in money if the bug is in a production version and you have marketing and patching costs in addition to the cost of finding and fixing a bug.

We have a card class that can display any card, so what we need next is a deck of cards. This deck needs to be able to be shuffled and we need to be able to deal cards from the deck. So, how would we test such a beast?

In a sorted deck, the cards should be from 1 to 52 so simply drawing cards from a sorted deck and making sure they are in sequence would solve this problem. A shuffled deck, however, has two issues that we need to check for. First, we want to make sure that all the cards are in the shuffle (that we haven’t duplicated cards losing some cards or other similar problem). Next we want tom make sure the cards are not in order. Simply checking to see the card is different from its position in the deck will not work as we would expect that at least some of the time a randomly shuffled card would end up in the proper deck position. A good shuffle would have at least half the cards in a different position so simply counting how many cards don’t match would suffice most of the time.

As the deck class can be written without the card class, as it uses card IDs and not instances of cards, it is possible to write the deck class in its own html file which I called deckTest. This starts off with your typical html5 boilerplate followed by a simple version of the Deck class with none of the necessary methods implemented.

Deck JavaScript Test

test results go here

Now some of you may have noticed that there is a lot of test code here. This is the downside to test driven development. You will often end up with test code that is the same size as the code that you are testing, and sometimes even larger. The plus side is that you know your code works within the testing parameters and if you refactor or change code you will be able to check to make sure that nothing you have done is broken. Manual testing could have been done with a lot less code. As always there is a trade-off that must be made. In a solo environment, and possibly a small team, manual testing may be adequate and may be productive enough. As you start to get to larger teams, people inadvertently mucking up other people’s code happens a lot more often than you would expect.

No comments: