Google Testing Blog

Testing on the Toilet: Only Verify State-Changing Method Calls

Monday, December 11, 2017

This article was adapted from a Google Testing on the Toilet (TotT) episode. You can download a printer-friendly version of this TotT episode and post it in your office.
By Dillon Bly and Andrew Trenk
Which lines can be safely removed from this test?

@Test public void addPermissionToDatabase() {
  new UserAuthorizer(mockUserService, mockPermissionDb).grantPermission(USER, READ_ACCESS);

  // The test will fail if any of these methods is not called.
  verify(mockUserService).isUserActive(USER);
  verify(mockPermissionDb).getPermissions(USER);
  verify(mockPermissionDb).isValidPermission(READ_ACCESS);
  verify(mockPermissionDb).addPermission(USER, READ_ACCESS);
}

The answer is that the calls to verify the non-state-changing methods can be removed.
Method calls on another object fall into one of two categories:

State-changing: methods that have side effects and change the world outside the code under test, e.g., sendEmail(), saveRecord(), logAccess().

Non-state-changing: methods that return information about the world outside the code under test and don't modify anything, e.g., getUser(), findResults(), readFile().

You should usually avoid verifying that non-state-changing methods are called:

It is often redundant: a method call that doesn't change the state of the world is meaningless on its own. The code under test will use the return value of the method call to do other work that you can assert.
It makes tests brittle: tests need to be updated whenever method calls change. For example, if a test is expecting mockUserService.isUserActive(USER) to be called, it would fail if the code under test is modified to call user.isActive() instead.
It makes tests less readable: the additional assertions in the test make it more difficult to determine which method calls actually affect the state of the world.
It gives a false sense of security: just because the code under test called a method does not mean the code under test did the right thing with the method’s return value.

Instead of verifying that they are called, use non-state-changing methods to simulate different conditions in tests, e.g., when(mockUserService.isUserActive(USER)).thenReturn(false). Then write assertions for the return value of the code under test, or verify state-changing method calls. Verifying non-state-changing method calls may be useful if there is no other output you can assert. For example, if your code should be caching an RPC result, you can verify that the method that makes the RPC is called only once. With the unnecessary verifications removed, the test looks like this:

@Test public void addPermissionToDatabase() {
  new UserAuthorizer(mockUserService, mockPermissionDb).grantPermission(USER, READ_ACCESS);

  // Verify only the state-changing method.
  verify(mockPermissionDb).addPermission(USER, READ_ACCESS);
}

That’s much simpler! But remember that instead of using a mock to verify that a method was called, it would be even better to use a real or fake object to actually execute the method and check that it works properly. For example, the above test could use a fake database to check that the permission exists in the database rather than just verifying that addPermission() was called.

You can learn more about this topic in the book Growing Object-Oriented Software, Guided by Tests. Note that the book uses the terms “command” and “query” instead of “state-changing” and “non-state-changing”.

Code Health: Obsessed With Primitives?

Monday, November 13, 2017

This is another post in our Code Health series. A version of this post originally appeared in Google bathrooms worldwide as a Google Testing on the Toilet episode. You can download a printer-friendly version to display in your office.
By Marc Eaddy
Programming languages provide basic types, such as integers, strings, and maps, which are useful in many contexts. For example, a string can be used to hold everything from a person's name to a web page URL. However, code that relies too heavily on basic types instead of custom abstractions can be hard to understand and maintain.
Primitive obsession is the overuse of basic ("primitive") types to represent higher-level concepts. For example, this code uses basic types to represent shapes:

vector<pair<int, int>> polygon = ...
pair<pair<int, int>, pair<int, int>> bounding_box = GetBoundingBox(polygon);
int area = (bounding_box.second.first  - bounding_box.first.first) *
           (bounding_box.second.second - bounding_box.first.seco

pair is not the right level of abstraction because its generically-named first and second fields are used to represent X and Y in one case and lower-left (er, upper-left?) and upper-right (er, lower-right?) in the other. Worse, basic types don't encapsulate domain-specific code such as computing the bounding box and area.Replacing basic types with higher-level abstractions results in clearer and better encapsulated code:

Polygon polygon = ...
int area = polygon.GetBoundingBox().GetArea();

Here are some other examples of primitive obsession:

Related maps, lists, vectors, etc. that can be easily combined into a single collection by consolidating the values into a custom higher-level abstraction.
map<UserId, string> id_to_name; map<UserId, int> id_to_age;
map<UserId, Person> id_to_person;
A vector or map with magic indices/keys, e.g. string values at indices/keys 0, 1, and 2 hold name, address, and phone #, respectively. Instead, consolidate these values into a higher-level abstraction.
person_data[kName] = "Foo";
person.SetName("Foo");
A string that holds complex or structured text (e.g. a date). Instead, use a higher-level abstraction (e.g. Date) that provides self-documenting accessors (e.g. GetMonth ) and guarantees correctness.
string date = "01-02-03";
Date date(Month::Feb, Day(1), Year(2003));
An integer or floating point number that stores a time value, e.g. seconds. Instead, use a structured timestamp or duration type.
int timeout_secs = 5;
Duration timeout = Seconds(5);

It's possible for any type—from a lowly int to a sophisticated red-black tree—to be too primitive for the job. If you see code that uses a lot of basic types that would be clearer or better encapsulated by using a higher-level abstraction, refactor it or politely remind the author to keep it classy!

Code Health: IdentifierNamingPostForWorldWideWebBlog

Monday, October 23, 2017

This is another post in our Code Health series. A version of this post originally appeared in Google bathrooms worldwide as a Google Testing on the Toilet episode. You can download a printer-friendly version to display in your office.
By Chris Lewis and Bob Nystrom

It's easy to get carried away creating long identifiers. Longer names often make things more readable. But names that are too long can decrease readability. There are many examples of variable names longer than 60 characters on GitHub and elsewhere. In 58 characters, we managed this haiku for you to consider:
Name variables Using these simple guidelines Beautiful source code
Names should be two things: clear (know what it refers to) and precise (know what it does not refer to). Here are some guidelines to help:

• Omit words that are obvious given a variable's type declaration.

// Bad, the type tells us what these variables are:
String nameString; List<datetime> holidayDateList;
// Better:
String name; List<datetime> holidays;

• Omit irrelevant details.

// Overly specific names are hard to read:
Monster finalBattleMostDangerousBossMonster; Payments nonTypicalMonthlyPayments;
// Better, if there's no other monsters or payments that need disambiguation:
Monster boss; Payments payments;

• Omit words that are clear from the surrounding context.

// Bad, repeating the context:
class AnnualHolidaySale {int annualSaleRebate; boolean promoteHolidaySale() {...}}
// Better:
class AnnualHolidaySale {int rebate; boolean promote() {...}}

• Omit words that could apply to any identifier.You know the usual suspects: data, state, amount, number, value, manager, engine, object, entity, instance, helper, util, broker, metadata, process, handle, context. Cut them out.
There are some exceptions to these rules; use your judgment. Names that are too long are still better than names that are too short. However, following these guidelines, your code will remain unambiguous and be much easier to read. Readers, including "future you,” will appreciate how clear your code is!

Code Health: Providing Context with Commit Messages and Bug Reports

Monday, September 11, 2017

This is another post in our Code Health series. A version of this post originally appeared in Google bathrooms worldwide as a Google Testing on the Toilet episode. You can download a printer-friendly version to display in your office.
By Chris Lewis
You are caught in a trap. Mechanisms whirl around you, but they make no sense. You desperately search the room and find the builder's original plans! The description of the work order that implemented the trap reads, "Miscellaneous fixes." Oh dear.
Reading other engineers' code can sometimes feel like an archaeology expedition, full of weird and wonderful statements that are hard to decipher. Code is always written with a purpose, but sometimes that purpose is not clear in the code itself. You can address this knowledge gap by documenting the context that explains why a change was needed. Code comments provide context, but comments alone sometimes can’t provide enough.
There are two key ways to indicate context:Commit Messages

A commit message is one of the easiest, most discoverable means of providing context. When you encounter lines of code that may be unclear, checking the commit message which introduced the code is a great way to gain more insight into what the code is meant to do.

Write the first line of the commit message so it stands alone, as tools like GitHub will display this line in commit listing pages. Stand-alone first lines allow you to skim through code history much faster, quickly building up your understanding of how a source file evolved over time. Example:

Add Frobber to the list of available widgets.

This allows consumers to easily discover the new Frobber widget and
add it to their application.

Bug Reports

You can use a bug report to track the entire story of a bug/feature/refactoring, adding context such as the original problem description, the design discussions between the team, and the commits that are used to solve the problem. This lets you easily see all related commits in one place, and allows others to easily keep track of the status of a particular problem.
Most commits should reference a bug report. Standalone commits (e.g. one-time cleanups or other small unplanned changes) don't need their own bug report, though, since they often contain all their context within the description and the source changes.

Informative commit messages and bug reports go hand-in-hand, providing context from different perspectives. Keep in mind that such context can be useful even to yourself, providing an easy reminder about the work you did last week, last quarter, or even last year. Future you will thank past you!

Code Health: Eliminate YAGNI Smells

Monday, August 14, 2017

This is another post in our Code Health series. A version of this post originally appeared in Google bathrooms worldwide as a Google Testing on the Toilet episode. You can download a printer-friendly version to display in your office.
By Marc Eaddy
The majority of software development costs are due to maintenance. One way to reduce maintenance costs is to implement something only when you actually need it, a.k.a. the “You Aren't Gonna Need It” (YAGNI) design principle. How do you spot unnecessary code? Follow your nose!
A code smell is a code pattern that usually indicates a design flaw. For example, creating a base class or interface with only one subclass may indicate a speculation that more subclasses will be needed in the future. Instead, practice incremental development and design: don't add the second subclass until it is actually needed.
The following C++ code has many YAGNI smells:

class Mammal { ...
  virtual Status Sleep(bool hibernate) = 0;
};
class Human : public Mammal { ...
  virtual Status Sleep(bool hibernate) {
    age += hibernate ? kSevenMonths : kSevenHours;
    return OK;
  }
};

Maintainers are burdened with understanding, documenting, and testing both classes when only one is really needed. Code must handle the case when hibernate is true, even though all callers pass false, as well as the case when Sleep returns an error, even though that never happens. This results in unnecessary code that never executes. Eliminating those smells simplifies the code:

class Human { ...
  void Sleep() { age += kSevenHours; }
};

Here are some other YAGNI smells:

Code that has never been executed other than by tests (a.k.a. code that is dead on arrival)
Classes designed to be subclassed (have virtual methods and/or protected members) that are not actually subclassed
Public or protected methods or fields that could be private
Parameters, variables, or flags that always have the same value

Thankfully, YAGNI smells, and code smells in general, are often easy to spot by looking for simple patterns and are easy to eliminate using simple refactorings.

Are you thinking of adding code that won't be used today? Trust me, you aren't gonna need it!

Testing Blog

Testing on the Toilet: Only Verify State-Changing Method Calls

Code Health: Obsessed With Primitives?

Code Health: IdentifierNamingPostForWorldWideWebBlog

Code Health: Providing Context with Commit Messages and Bug Reports

Code Health: Eliminate YAGNI Smells

Labels

Archive

Feed

Subscribe by Email

Company-wide

Products

Developers