Transcript Refactoring 4

Refactoring IV
26-Jul-16
Previously discussed bad smells










Duplicated code — and other forms of redundancy
Long method — use short methods that delegate work
Large class — trying to do too much
Long parameter list — hard to use and remember
Divergent change — changes in one class for different reasons
Shotgun surgery — a change requires little changes all over
Feature envy — method uses too much from some other class
Data clumps — variables that frequently occur together
Primitive obsession — being afraid of making “small” objects
Switch statements — probably should use polymorphism
instead
More bad smells












Parallel inheritance hierarchies — can’t make just one subclass
Lazy class — too few responsibilities
Speculative generality — code that isn’t needed
Temporary field — an object doesn’t use all its variables
Message chains — asking for objects to ask for objects
Middle man — too much responsibility passed along
Inappropriate intimacy — classes accessing each other too much
Alternative classes with different interfaces — similar work but
with different signatures
Incomplete library class — inadequate for reuse
Data class — just data, maybe getters and setters
Refused bequest — subclass don’t use much of their inheritance
Comments — when used as a substitute for good code
Parallel inheritance hierarchies


When you make a subclass of one class, you have to
make a corresponding subclass of another class
General strategy: Use Move Method and Move Field to
make instances of one hierarchy refer to instances of the
other
Lazy class

Small classes are fine, but sometimes a class just
doesn’t do enough

If a class is very similar to its superclass, you can try to use
Collapse Hierarchy to merge the two classes



Eliminate the subclass by using Pull Up Field and Pull Up Method; or,
Eliminate the superclass by using Push Down Field and Push Down
Method
If a class just isn’t doing very much, move all its features into
another class with Move Field and Move Method
Speculative generality

One of the principles of Extreme Programming (XP) is
that you shouldn’t write code until you need it



XP assumes that code will change frequently, and tries to
make change as fast and easy as possible
If you try to make things too general, you may have
unnecessary code that just gets in the way
If the only users of a class or method are test cases, the
code should be thrown away

Since I hate to throw away good code, I usually move unused
code to a discards directory
Temporary field


We expect an object to use all its fields
It’s confusing when an instance variable is used only in
certain cases



Use Extract Class to create a home for these variables
Eliminate conditional code with Introduce Null Object
Sometimes programmers will add instance variables to
avoid long parameter lists between communicating
methods

Use Extract Class to create a new method object
Message chains

A message chain is a sequence such as
BazObject b =foo.getBar().getBaz()





Here we are asking foo for a bar object so that we can ask it
for a baz object
Often, but not always, these are getter methods
Message chains can be abbreviated or eliminated by Hide
Delegate
As an example of Hide Delegate, we can introduce the
following method into foo’s class:
 BazObject getBaz() { return bar.getBaz(); }
And then we can just call BazObject b = foo.getBaz();
Middle man

Delegation—providing methods to call methods in another
class—is often useful for hiding internal details



Example:
BazObject getBaz() { return bar.getBaz(); }
However, too much delegation isn’t good
You can:



Use Remove Middle Man and talk to the object that really knows what is
going on
Use Inline Method to absorb a few small methods into the caller
Use Replace Delegation With Inheritance to turn the middle man into a
subclass of the real object
Inappropriate intimacy

Classes may make too much use of each other’s fields
and methods


Use Move Method and Move Field to reduce the association
Try to Change Bidirectional Association to Unidirectional




The idea here is to take the class that is less dependent on the other
class and remove the remaining dependencies
If the classes have common needs, try Extract Class
Use Hide Delegate to let another class act as a middle man
If a subclass knows too much about its superclass, use
Replace Inheritance With Delegation (see later comments on
Refused Bequest)
Replace Inheritance With Delegation



Sometimes a subclass inherits more from its superclass than you
want it to have
Example: Suppose class Sub extends Super, inherits desired
methods int foo() and void bar(), along with other methods it
does not want, and adds method int baz()
Replace
class Sub extends Super {...}
with
class Sub { // class name should also be changed
Super s = new Super();
int foo() { return s.foo(); } // delegate to s
void bar() { s.bar(); }
// delegate to s
int baz() {...} // new method
}
Alternative classes, different interfaces

You end up with two essentially equivalent classes
(example: Java’s Enumeration and Iterator classes)





Java can’t eliminate Enumeration because that would break
old code
Even in this situation, the functionality can be moved into a
single class
Use Rename Method on methods that do the same thing
but have different signatures
Use Move Method until classes are doing the same
things
You may want to use Extract Superclass
Incomplete library class

Library classes (such as those supplied by Sun) don’t always do
everything we want them to do


Use Introduce Foreign Method:




It’s usually impossible to modify these library classes
Write the method you want, as if it were in the library class
Make an instance of the library class the first parameter
Add a comment that describes the method as a “foreign method”
Example:

private static Date nextDay(Date arg) {
// foreign method, should be in Date
return new Date(arg.getYear(), arg.getMonth(),
arg getDate() + 1);
}
Data class



Classes that just hold data, and maybe setters and getters for that
data, are undesirable
If a field is public, use Encapsulate Method to make it private and
add setters and getters (if it isn’t too late)
If a method returns a collection, use Encapsulate Collection



Make the method return a read-only view (java.util.Collections supplies
methods such as unmodifiableSet(Set) and unmodifiableMap(Map))
Provide add and remove methods as appropriate
Try to use Move Method to move the setters and getters to the
class that needs them
Refused bequest

Subclasses may inherit unwanted methods from their
superclasses

This suggests that the hierarchy may be wrong


Create a new subclass and use Push Down Method and Push Down
Field on the unused methods
Fowler feels that this isn’t a severe problem, unless the
subclass is reusing behavior but not the interface


In a few places, Sun’s packages override an inherited method with one
that just throws an UnsupportedOperationException
A better solution is to Replace Inheritance With Delegation
Comments

Fowler says “comments often are used as a deodorant”




If you need a comment to explain what a block of code does,
use Extract Method
If you need a comment to explain what a method does, use
Rename Method
If you need to describe the required state of the system, use
Introduce Assertion
This should not discourage the use of comments
(especially javadoc comments)—the point is that code
should be self-explanatory, so that comments are not
necessary
Advice from Kent Beck

“Stopping is the strongest move in the refactorer’s repertoire.”



“Somewhere your code smells bad. Resolve to get rid of the
problem. Then march toward that goal.”
Stop when you are unsure



If the code is already better, go ahead and release it
If you aren’t sure of your changes, discard them and go back to the
previous stable state
Backtrack. If the code quits working, and you haven’t tested in a
while, don’t debug—throw away the changes since the last
working tests


The idea is to take small goals, refactor in small steps, so that you can quit
in a stable state, having made progress
Emotionally difficult, but rewriting is faster and easier than debugging
Work with someone. There are many advantages to working in
pairs.
Soundbites I

When you find you have to add a feature to a program,
and the program’s code is not structured in a convenient
way to add the feature, first refactor the program to
make it easy to add the feature, then add the feature

Before you start refactoring, check that you have a solid
suite of tests. These tests must be self-checking.

Refactoring changes the programs in small steps. If you
make a mistake, it is easy to find the bug.
Soundbites II

Any fool can write code that a computer can understand. Good
programmers write code that humans can understand.

Refactoring (noun): a change made to the internal structure of
software to make it easier to understand and cheaper to modify
without changing the observable behavior of the software.

Refactor (verb): to restructure software by applying a series of
refactorings without changing the observable behavior of the
software.
Soundbites III

Three strikes and you refactor.



“The first time you do something, you just do it. The second
time you do something similar, you wince at the duplication,
but you do the duplicate thing anyway. The third time you do
something similar, you refactor.”
Don’t publish interfaces prematurely. Modify your code
ownership policies to smooth refactoring.
When you feel the need to write a comment, first try to
refactor the code so that any comment becomes
superfluous.
Soundbites IV

Make sure all tests are fully automatic and that they
check their own results.

A suite of tests is a powerful bug detector that
decapitates the time it takes to find bugs.

Run your tests frequently. Localize tests whenever you
compile—every test at least every day.

When you get a bug report, start by writing a unit test
that exposes the bug.
Soundbites V

It is better to write and run incomplete tests than not to
run complete tests.

Think of the boundary conditions under which things
might go wrong and concentrate your tests there.


Don’t forget to test that exceptions are raised when
things are expected to go wrong.
Don’t let the fear that testing can’t catch all bugs stop
you from writing the tests that will catch most bugs.
Conclusions

A couple of the more important goals of Agile
Programming (including XP) are to:




Keep code flexible and easily changed
Avoid long debugging sessions
Always have something that works
Refactoring depends heavily on:


Taking small steps
Having a fully automated test suite that you use frequently
The End