The Test

As I poured another cup of coffee, my eyelids hung, so I was able to stay awake enough to study the last few thousand pages of my notebook. Perhaps there are too many caffeine, but I say that "You do not have hope, sleep," I think the coffee maker is laughing at me. Things are beginning to confuse me. Facts and figures are mixed in many of my minds, and useless dates and numbers are meaningless to me in my exhausted condition.

Our project contains many tests. There are autonomous test suites that can run unit tests, sustainability tests and integration tests. Unit testing takes a few minutes to complete, but all of them take an hour! I noticed that quick testing was the best and I do not need to keep using outdated old tests. We use Subversion for version management. Unfortunately, the code we send is automatically checked into the repository. Opportunity cost seems to be too high, so we rarely cooperate with affiliates. This reduces the frequency with which the code is sent. I am still trying to submit it frequently, but I interrupt the build sometimes - I think that it is not necessary to spend time to run tests locally

Everything you need for building, deploying, testing, and publishing needs to be stored in some form of version store. This includes requirement documents, test scripts, test cases, libraries, tool chains, and so on. All these things are versioned and should be able to identify the given build. Please do not leave memories to human beings. Even the best human beings are notoriously bad. The famous motto of "building quality", Edward Deming should be a guide for everyone who works on the project. The faster you find defects, the lower their repair costs and the less likely they will fall on the customer's desktop. Previously, I talked about "bringing pain". The next step is to fix the pain immediately after seeing the pain.

Do not leave unsuccessful tests unattended - everyone knows that tests may fail. Modifying specific tests is a difficult task and sometimes fails (although we promise to fix it later). The result is usually the same. The setting changes to red, everyone will ignore the newly failed test with this setting, eventually the effect of the test will be gone. Therefore, the test is executed, but the result is usually ignored. To solve this problem, the team needs some training. If the test fails and is temporary, you can skip these tests (that is, throw SkipException in your code) or you can use the tool to mute the test. Therefore, it does not affect the actual situation. Do not forget the appropriate comment