Naresh Jain » Agile FAQs Blog » Managed Chaos » My Test Naming Conventions: More Examples

My Test Naming Conventions: More Examples

Other’s Naming Convention	My Naming Convention
itShouldSetAccountToLoggedInWhen- PasswordMatches	matchingPasswordLogsYouIn
itShouldSetAccountToRevokedAfterThree- FailedLoginAttempts	threeFailedLoginAttemptsRevokesYourAccount
itShouldNotSetAccountLoggedInIf- PasswordDoesNotMatch	nonMatchingPasswordDoesNotLogYouIn
itShouldNotRevokeSecondAccountAfterTwo- FailedAttemptsFirstAccount	failedLoginAttemptsOnOtherAccountsDoes- NotImpactMyAccount
itShouldNowAllowConcurrentLogins	sameUserCannotLoginConcurrently
itShouldThrowExceptionIfAccountNotFound	onlyUserWithExistingAccountCanLogin
ItShouldNotBePossibleToLogIntoRevokedAccount	userWithRevokedAccountCannotLogin
itShouldResetBackToInitialStateAfter- SuccessfulLogin	successfulLoginResetsFailedAttemptCount

So why do I prefer calling my test as nonMatchingPasswordDoesNotLogYouIn instead of itShouldSetAccountToLoggedInWhenPasswordMatches?

I think the latter is focusing on implementation details versus focusing on the intent.
One should write descriptive names, but the longer the name is, that much more difficult it is to read. I want to write my test names such that they are short and concise. Hence I think the whole itShould<> thing seems repetitive without adding much value.

More about my test naming convention here…

This entry was posted on Sunday, June 14th, 2009 at 9:38 pm and is filed under Design, Programming, Testing. You can follow any responses to this entry through the RSS 2.0 feed. You can skip to the end and leave a response. Pinging is currently not allowed.

Sai
An important question is how you refer your test class names. I prefer to names them as same as production class names. Example (taking yours to avoid confusion) if you are writing a mail app (AcmeMail) so I would name test as
```
AcmeMail -> reset_failed_attempt_count_if_I_successfully_log_in
```
.
```
AcmeMail -> donot_allow_login_if_my_account_is_revoked
```
Yes I do use _ instead of camel casing. Is it too extreme? :)

Sai
okay... The font for the code isn't great. If you use a electron microscope it may be visible as a nano particle in it. One more try and I promise this is the last one
```
AcmeMail -> reset_failed_attempt_count_if_I_successfully_log_in
```
.
```
AcmeMail -> donot_allow_login_if_my_account_is_revoked
```

Naresh Jain

@Sai, I don't mind the underscore. They can make the code more readable. However I find that typing an underscore is a bit more painful than typing a capital letter. Hence I prefer sticking with the Camel Case.

Blaise Pascal

So why do I prefer calling my test as nonMatchingPasswordDoesNotLogYouIn instead of itShouldSetAccountToLoggedInWhenPasswordMatches?

I find your reasons reasonable, but incomplete. Why would you rename itShouldSetAccountToLoggedInWhenPasswordMatches to nonMatchingPasswordDoesNotLogYouIn instead of matchingPasswordLogsYouIn?

Sai

Well the question was not just for camel case (its a carry forward from ruby) also for naming the test class and test method names.

So if others name as itShouldSetAccountToLoggedInWhenPasswordMatches and you as nonMatchingPasswordDoesNotLogYouIn, I tend to name it as AcmeMail -> logs_you_in_when_passwords_when_passwords_match. The name tends to be descriptive and doesn't have implementation details. I see logging in when password matches and not logging in when password don't match as two different things.

Also I agree with Blaise Pascal (Is it really Pascal? :)

Blaise Pascal

Sai:

I've been using the name Blaise Pascal online for over a quarter century. It's not what it says on my legal documentation, but it's real enough.

Naresh Jain

@Pascal, Good catch. That was a Ctrl+C Ctrl+V error. Fixed it now.

@Sai, very interesting. I've never used the same test class name as the production class name. If I were to do so, I would have to use a different namespace/package. I'm old school and tend to put my test classes in the same package as the production classes. The main reason I do so is, I usually have only one public class in a package public and rest of the classes are package protected (package level visibility). So when I drive their implementation by writing tests, I need to put my tests in the same package.

There are 2 main styles I use when it comes to naming tests:
<ul>
<li><prodClassName>Test - Standard xUnit convention </li>
<li>Behavior (verbs) as Test Class Names</li>
</ul>

Agile India.

My Test Naming Conventions: More Examples