We Keep Coding

Should I strictly avoid using enums on Android?

I used to define a set of related constants like Bundle keys together in an interface like below:
public interface From{
String LOGIN_SCREEN = "LoginSCreen";
String NOTIFICATION = "Notification";
String WIDGET = "widget";
}
This provides me a nicer way to group related constants together and used them by making a static import (not implements). I know Android framework also uses the constants in same way like Toast.LENTH_LONG, View.GONE.
However, I often feel that the Java Enums provide much better and powerful way to represent the constant.
But is there a performence issue in using enums on Android?
With a bit of research I ended up in confusion. From this question
"Avoid Enums Where You Only Need Ints” removed from Android's performance tips? it's clear that Google has removed "Avoid enums" from its performance tips, but from it's official training docs Be aware of memory overhead section it clearly says: "Enums often require more than twice as much memory as static constants. You should strictly avoid using enums on Android." Is this still holds good? (say in Java versions after 1.6)
One more issue that I observed is to send enums across intents using Bundle I should send them by serializing (i.e putSerializable(), that I think an expensive operation compared to primitive putString() method, eventhough enums provides it for free).
Can someone please clarify which one is the best way to represent the same in Android? Should I strictly avoid using enums on Android?

Use enum when you need its features. Don't avoid it strictly.
Java enum is more powerful, but if you don't need its features, use constants, they occupy less space and they can be primitive itself.
When to use enum:
type checking - you can accept only listed values, and they are not continuous (see below what I call continuous here)
method overloading - every enum constant has its own implementation of a method
public enum UnitConverter{
METERS{
#Override
public double toMiles(final double meters){
return meters * 0.00062137D;
}
#Override
public double toMeters(final double meters){
return meters;
}
},
MILES{
#Override
public double toMiles(final double miles){
return miles;
}
#Override
public double toMeters(final double miles){
return miles / 0.00062137D;
}
};
public abstract double toMiles(double unit);
public abstract double toMeters(double unit);
}
more data - your one constant contains more than one information that cannot be put in one variable
complicated data - your constant need methods to operate on the data
When not to use enum:
you can accept all values of one type, and your constants contain only these most used
you can accept continuous data
public class Month{
public static final int JANUARY = 1;
public static final int FEBRUARY = 2;
public static final int MARCH = 3;
...
public static String getName(final int month){
if(month <= 0 || month > 12){
throw new IllegalArgumentException("Invalid month number: " + month);
}
...
}
}
for names (like in your example)
for everything else that really doesn't need an enum
Enums occupy more space
a single reference to an enum constant occupies 4 bytes
every enum constant occupies space that is a sum of its fields' sizes aligned to 8 bytes + overhead of the object
the enum class itself occupies some space
Constants occupy less space
a constant doesn't have a reference so it's a pure data (even if it's a reference, then enum instance would be a reference to another reference)
constants may be added to an existing class - it's not necessary to add another class
constants may be inlined; it brings extended compile-time features (such as null checking, finding dead code etc.)

If the enums simply have values, you should try to use IntDef/StringDef , as shown here:
https://developer.android.com/studio/write/annotations.html#enum-annotations
Example: instead of :
enum NavigationMode {NAVIGATION_MODE_STANDARD, NAVIGATION_MODE_LIST, NAVIGATION_MODE_TABS}
you use:
#IntDef({NAVIGATION_MODE_STANDARD, NAVIGATION_MODE_LIST, NAVIGATION_MODE_TABS})
#Retention(RetentionPolicy.SOURCE)
public #interface NavigationMode {}
public static final int NAVIGATION_MODE_STANDARD = 0;
public static final int NAVIGATION_MODE_LIST = 1;
public static final int NAVIGATION_MODE_TABS = 2;
and in the function that has it as a parameter/returned value , use:
#NavigationMode
public abstract int getNavigationMode();
public abstract void setNavigationMode(#NavigationMode int mode);
In case the enum is complex, use an enum. It's not that bad.
To compare enums vs constant values, you should read here:
http://hsc.com/Blog/Best-Practices-For-Memory-Optimization-on-Android-1
Their example is of an enum with 2 values. It takes 1112 bytes in dex file compared to 128 bytes when constant integers are used . Makes sense, as enums are real classes, as opposed to how it works on C/C++ .

With Android P, google has no restriction/objection in using enums
The documentation has changed where before it was recommended to be cautious but it doesn't mention it now.
https://developer.android.com/reference/java/lang/Enum

In addition to previous answers, I would add that if you are using Proguard (and you should definitely do it to reduce size and obfuscate your code), then your Enums will be automatically converted to #IntDef wherever it is possible:
https://www.guardsquare.com/en/proguard/manual/optimizations
class/unboxing/enum
Simplifies enum types to integer constants, whenever possible.
Therefore, if you have some discrete values and some method should allow to take only this values and not others of the same type, then I would use Enum, because Proguard will make this manual work of optimizing code for me.
And here is a good post about using enums from Jake Wharton, take a look at it.
As a library developer, I recognize these small optimizations that should be done as we want to have as little impact on the consuming app's size, memory, and performance as possible. But it's important to realize that [...] putting an enum in your public API vs. integer values where appropriate is perfectly fine. Knowing the difference to make informed decisions is what's important

Should I strictly avoid using enums on Android?
No. "Strictly" means they are so bad, they should not be used at all. Possibly a performance issues might arise in an extreme situation like many many many (thousands or millions of) operations with enums (consecutive on the ui thread). Far more common are the network I/O operations that should strictly happen in a background thread.
The most common usage of enums is probably some kind of type check - whether an object is this or that which is so fast you won't be able to notice a difference between a single comparison of enums and a comparison of integers.
Can someone please clarify which one is the best way to represent the same in Android?
There is no general rule of thumb for this. Use whatever works for you and helps you get your app ready. Optimize later - after you notice there's a bottleneck that slows some aspect of your app.

I like to add, that you can not use #Annotations when you declare a List<> or Map<> where either key or value is of one of your annotation interfaces.
You get the error "Annotations are not allowed here".
enum Values { One, Two, Three }
Map<String, Values> myMap; // This works
// ... but ...
public static final int ONE = 1;
public static final int TWO = 2;
public static final int THREE = 3;
#Retention(RetentionPolicy.SOURCE)
#IntDef({ONE, TWO, THREE})
public #interface Values {}
Map<String, #Values Integer> myMap; // *** ERROR ***
So when you need to pack it into a list/map, use enum, as they can be added, but #annotated int/string groups can not.

Two facts.
1, Enum is one of the most powerful feature in JAVA.
2, Android phone usually has a LOT of memory.
So my answer is NO. I will use Enum in Android.

private static final vs private final

This question has been asked here. As per the answer :
private final int NUMBER = 10;
If it cannot change, there is no point having one copy per instance.
My doubt is what if instance of the class is created, say once a day and it lasts around a few seconds. Is it good idea to keep the int(in some case object) in memory?
Assuming, there can be many (20-30) such objects.

How you store the information depends very much on what it is intended to be used for.
There are a few approaches you might take:
private static final
This is a good choice if the value will never be modified during the lifetime of the application. It means, when you're creating your multiple instances, you are only actually storing this particular variable ONCE.
private final
This is meant for those times when the value might take on different values for different instances of your object, but any specific instance will not have it's value modified throughout the object's life time.
If you're looking at something which might take on different values over a range of time, then this might be of interest to you.
public static int GetNumber(){...}
Another approach you might consider is to have a static method return the value you are after. This makes it easy to deal with changes in the value, but you also need to consider the effect of such a change throughout the lifetime of any given instance.
Hope that helps...

Regarding private final int number, your claim that
If it cannot change, there is no point having one copy per instance
is absolutely wrong. Consider this code, typical of an immutable class:
private final int number;
public MyClass(int number) {
this.number = number;
}
There are many instances of this pattern in the JDK, and in production code classes around the globe.
To answer your question about performance, I doubt you could measure the difference between using static vs instance constants.

Will things run quicker if I make my variables final?

I'm writing for Android (Java).
I'm declaring int's and float's as part of an ongoing loop.
Some of them don't need to be changed after declaration.
If I set them all to final when declaring, will things run quicker?
[Edit]
Thanks everyone. I didn't actually expect it to make any improvements, I just noticed, after browsing the source of various large projects, it was fairly common. Cheers

Things will not run quicker. The final keyword is just compile time syntactic sugar.
If it were actually static final, then you could take benefit of compiletime calculation and inlining of the value in any refernce. So, with for example:
private static final long ONE_WEEK_IN_MILLIS = 7 * 24 * 60 * 60 * 1000L;
public void foo(Date date) {
if (date.getTiem() > System.currentTimeMillis() + ONE_WEEK_IN_MILLIS) {
// No idea what to do here?
}
}
the compiler will optimize one and other so that it ends up like:
private static final long ONE_WEEK_IN_MILLIS = 604800000L;
public void foo(Date date) {
if (date.getTiem() > System.currentTimeMillis() + 604800000L) {
// No idea what to do here?
}
}
If you run a decompiler, you'll see it yourself.

Although setting to final might have impact on the speed, the answer will most probably be different for each VM or device.
Declaring them final, however, doesn't hurt, and one could even call it good programming style.
As for performance, this looks almost certainly like premature optimization. Profile, find bottlenecks, rethink your algorithms. Don't waste your time with "final" just because of performance - it will barely solve any problem.

If you also make it static (a class variable) it can increase performance, and it is also good programming practice to use final for variables that you know will not change. Though, you may not want it to be a class variable, in which case, I am not sure if can improves performance, but I think it may in many cases.
http://docs.sun.com/app/docs/doc/819-3681/6n5srlhjs?a=view
The dynamic compiler can perform some constant folding optimizations easily, when you declare constants as static final variables.
Declare method arguments final if they are not modified in the method. In general, declare all variables final if they are not modified after being initialized or set to some value.
So for example, if you have code that multiples two of your final variables, during run-time the VM may use what would normally be sleep/downtime to calculate the result of that multiplication so it doesn't have to do it in the busy periods.

I'd consider it a good practice to make variables final (you might use Eclipse's Preferences > Java > Code Style > Clean Up to do so). While performance might actually improve, I'd expect the differences to be negligible. In my opinion, it helps with readability of code though (i.e. no need to look for assignments) which certainly is a Good Thing (tm).

When we declare any variable final, means at compilation time it would be identified and while running the application JVM does not check it for any manipulation as it is declared as final(constant). so definately we are removind overhead from JVM.
so we could say it will improve performance, if depends on your case is the variable is constant make it final
better if you make if static final.....
they are optimized by JVM are kept in the Constant Pool with the Classfile "http://negev.wordpress.com/java-memory-brief/"

Simple, general-interest, code-analyzers based, Java questions [closed]

Closed. This question needs to be more focused. It is not currently accepting answers.
Want to improve this question? Update the question so it focuses on one problem only by editing this post.
Closed 7 years ago.
Improve this question
OK, after reviewing some code with PMD and FindBugs code analyzers, i was able to do great changes on the reviewed code. However, there are some things i don't know how to fix. I'll iterate them bellow, and (for better reference) i will give each question a number. Feel free to answer to any/all of them. Thanks for your patience.
1. Even tough i have removed some of the rules, the associated warnings are still there, after re-evaluate the code. Any idea why?
2. Please look at the declarations :
private Combo comboAdress;
private ProgressBar pBar;
and the references to objects by getters and setters :
private final Combo getComboAdress() {
return this.comboAdress;
}
private final void setComboAdress(final Combo comboAdress) {
this.comboAdress = comboAdress;
}
private final ProgressBar getpBar() {
return this.pBar;
}
private final void setpBar(final ProgressBar pBar) {
this.pBar = pBar;
}
Now, i wonder why the first declaration don't give me any warning on PMD, while the second gives me the following warning :
Found non-transient, non-static member. Please mark as transient or provide accessors.
More details on that warning here.
3. Here is another warning, also given by PMD :
A method should have only one exit point, and that should be the last statement in the method
More details on that warning here.
Now, i agree with that, but what if i write something like this :
public void actionPerformedOnModifyComboLocations() {
if (getMainTree().isFocusControl()) {
return;
}
....//do stuffs, based on the initial test
}
I tend to agree with the rule, but if performance of the code suggest multiple exit points, what should i do?
4. PMD gives me this :
Found 'DD'-anomaly for variable 'start_page' (lines '319'-'322').
when i declare something like :
String start_page = null;
I get rid of this info (level of warning is info) if i remove the assignment to null, but..i got an error from IDE, saying that the variable could be uninitialized, at some point later in the code. So, i am kind of stuck with that. Supressing the warning is the best you can do?
5. PMD Warning :
Assigning an Object to null is a code smell. Consider refactoring.
This is the case of a singletone use of GUI components or the case of a method who returns complex objects. Assigning the result to null in the catch() section it's justified by the need to avoid the return of an incomplete/inconsistent object. Yes, NullObject should be used, but there are cases where i don't want to do that. Should i supress that warning then?
6. FindBugs warning #1:
Write to static field MyClass.instance from instance method MyClass.handleEvent(Event)
in the method
#Override
public void handleEvent(Event e) {
switch (e.type) {
case SWT.Dispose: {
if (e.widget == getComposite()) {
MyClass.instance = null;
}
break;
}
}
}
of the static variable
private static MyClass instance = null;
The variable allows me to test whether the form is already created and visible or not, and i need to force the re-creation of the form, in some cases. I see no other alternative here. Any insights? (MyClass implements Listener, hence the overrided handleEvent() method).
7. FindBugs warning #2:
Class MyClass2 has a circular dependency with other classes
This warning is displayed based on simple imports of other classes. Do i need to refactor those imports to make this warning go away? Or the problem relies in MyClass2?
OK, enough said for now..expect an update, based on more findings and/or your answers. Thanks.

Here are my answers to some of your questions:
Question number 2:
I think you're not capitalizing the properties properly. The methods should be called getPBar and setPBar.
String pBar;
void setPBar(String str) {...}
String getPBar() { return pBar};
The JavaBeans specification states that:
For readable properties there will be a getter method to read the property value. For writable properties there will be a setter method to allow the property value to be updated. [...] Constructs a PropertyDescriptor for a property that follows the standard Java convention by having getFoo and setFoo accessor methods. Thus if the argument name is "fred", it will assume that the reader method is "getFred" and the writer method is "setFred". Note that the property name should start with a lower case character, which will be capitalized in the method names.
Question number 3:
I agree with the suggestion of the software you're using. For readability, only one exit point is better. For efficiency, using 'return;' might be better. My guess is that the compiler is smart enough to always pick the efficient alternative and I'll bet that the bytecode would be the same in both cases.
FURTHER EMPIRICAL INFORMATION
I did some tests and found out that the java compiler I'm using (javac 1.5.0_19 on Mac OS X 10.4) is not applying the optimization I expected.
I used the following class to test:
public abstract class Test{
public int singleReturn(){
int ret = 0;
if (cond1())
ret = 1;
else if (cond2())
ret = 2;
else if (cond3())
ret = 3;
return ret;
}
public int multReturn(){
if (cond1()) return 1;
else if (cond2()) return 2;
else if (cond3()) return 3;
else return 0;
}
protected abstract boolean cond1();
protected abstract boolean cond2();
protected abstract boolean cond3();
}
Then, I analyzed the bytecode and found that for multReturn() there are several 'ireturn' statements, while there is only one for singleReturn(). Moreover, the bytecode of singleReturn() also includes several goto to the return statement.
I tested both methods with very simple implementations of cond1, cond2 and cond3. I made sure that the three conditions where equally provable. I found out a consistent difference in time of 3% to 6%, in favor of multReturn(). In this case, since the operations are very simple, the impact of the multiple return is quite noticeable.
Then I tested both methods using a more complicated implementation of cond1, cond2 and cond3, in order to make the impact of the different return less evident. I was shocked by the result! Now multReturn() is consistently slower than singleReturn() (between 2% and 3%). I don't know what is causing this difference because the rest of the code should be equal.
I think these unexpected results are caused by the JIT compiler of the JVM.
Anyway, I stand by my initial intuition: the compiler (or the JIT) can optimize these kind of things and this frees the developer to focus on writing code that is easily readable and maintainable.
Question number 6:
You could call a class method from your instance method and leave that static method alter the class variable.
Then, your code look similar to the following:
public static void clearInstance() {
instance = null;
}
#Override
public void handleEvent(Event e) {
switch (e.type) {
case SWT.Dispose: {
if (e.widget == getComposite()) {
MyClass.clearInstance();
}
break;
}
}
}
This would cause the warning you described in 5, but there has to be some compromise, and in this case it's just a smell, not an error.
Question number 7:
This is simply a smell of a possible problem. It's not necessarily bad or wrong, and you cannot be sure just by using this tool.
If you've got a real problem, like dependencies between constructors, testing should show it.
A different, but related, problem are circular dependencies between jars: while classes with circular dependencies can be compiled, circular dependencies between jars cannot be handled in the JVM because of the way class loaders work.

I have no idea. It seems likely that whatever you did do, it was not what you were attempting to do!
Perhaps the declarations appear in a Serializable class but that the type (e.g. ComboProgress are not themselves serializable). If this is UI code, then that seems very likely. I would merely comment the class to indicate that it should not be serialized.
This is a valid warning. You can refactor your code thus:
public void actionPerformedOnModifyComboLocations() {
if (!getMainTree().isFocusControl()) {
....//do stuffs, based on the initial test
}
}
This is why I can't stand static analysis tools. A null assignment obviously leaves you open to NullPointerExceptions later. However, there are plenty of places where this is simply unavoidable (e.g. using try catch finally to do resource cleanup using a Closeable)
This also seems like a valid warning and your use of static access would probably be considered a code smell by most developers. Consider refactoring via using dependency-injection to inject the resource-tracker into the classes where you use the static at the moment.
If your class has unused imports then these should be removed. This might make the warnings disappear. On the other hand, if the imports are required, you may have a genuine circular dependency, which is something like this:
class A {
private B b;
}
class B {
private A a;
}
This is usually a confusing state of affairs and leaves you open to an initialization problem. For example, you may accidentally add some code in the initialization of A that requires its B instance to be initialized. If you add similar code into B, then the circular dependency would mean that your code was actually broken (i.e. you couldn't construct either an A or a B.
Again an illustration of why I really don't like static analysis tools - they usually just provide you with a bunch of false positives. The circular-dependent code may work perfectly well and be extremely well-documented.

For point 3, probably the majority of developers these days would say the single-return rule is simply flat wrong, and on average leads to worse code. Others see that it a written-down rule, with historical credentials, some code that breaks it is hard to read, and so not following it is simply wrong.
You seem to agree with the first camp, but lack the confidence to tell the tool to turn off that rule.
The thing to remember is it is an easy rule to code in any checking tool, and some people do want it. So it is pretty much always implemented by them.
Whereas few (if any) enforce the more subjective 'guard; body; return calculation;' pattern that generally produces the easiest-to-read and simplest code.
So if you are looking at producing good code, rather than simply avoiding the worst code, that is one rule you probably do want to turn off.

What's the best way to handle coexistence of the "int enum" pattern with java enums as an API evolves?

Suppose you're maintaining an API that was originally released years ago (before java gained enum support) and it defines a class with enumeration values as ints:
public class VitaminType {
public static final int RETINOL = 0;
public static final int THIAMIN = 1;
public static final int RIBOFLAVIN = 2;
}
Over the years the API has evolved and gained Java 5-specific features (generified interfaces, etc). Now you're about to add a new enumeration:
public enum NutrientType {
AMINO_ACID, SATURATED_FAT, UNSATURATED_FAT, CARBOHYDRATE;
}
The 'old style' int-enum pattern has no type safety, no possibility of adding behaviour or data, etc, but it's published and in use. I'm concerned that mixing two styles of enumeration is inconsistent for users of the API.
I see three possible approaches:
Give up and define the new enum (NutrientType in my fictitious example) as a series of ints like the VitaminType class. You get consistency but you're not taking advantage of type safety and other modern features.
Decide to live with an inconsistency in a published API: keep VitaminType around as is, and add NutrientType as an enum. Methods that take a VitaminType are still declared as taking an int, methods that take a NutrientType are declared as taking such.
Deprecate the VitaminType class and introduce a new VitaminType2 enum. Define the new NutrientType as an enum. Congratulations, for the next 2-3 years until you can kill the deprecated type, you're going to deal with deprecated versions of every single method that took a VitaminType as an int and adding a new foo(VitaminType2 v) version of each. You also need to write tests for each deprecated foo(int v) method as well as its corresponding foo(VitaminType2 v) method, so you just multiplied your QA effort.
What is the best approach?

How likely is it that the API consumers are going to confuse VitaminType with NutrientType? If it is unlikely, then maybe it is better to maintain API design consistency, especially if the user base is established and you want to minimize the delta of work/learning required by customers. If confusion is likely, then NutrientType should probably become an enum.
This needn't be a wholesale overnight change; for example, you could expose the old int values via the enum:
public enum Vitamin {
RETINOL(0), THIAMIN(1), RIBOFLAVIN(2);
private final int intValue;
Vitamin(int n) {
intValue = n;
}
public int getVitaminType() {
return intValue;
}
public static Vitamin asVitamin(int intValue) {
for (Vitamin vitamin : Vitamin.values()) {
if (intValue == vitamin.getVitaminType()) {
return vitamin;
}
}
throw new IllegalArgumentException();
}
}
/** Use foo.Vitamin instead */
#Deprecated
public class VitaminType {
public static final int RETINOL = Vitamin.RETINOL.getVitaminType();
public static final int THIAMIN = Vitamin.THIAMIN.getVitaminType();
public static final int RIBOFLAVIN = Vitamin.RIBOFLAVIN.getVitaminType();
}
This allows you to update the API and gives you some control over when to deprecate the old type and scheduling the switch-over in any code that relies on the old type internally.
Some care is required to keep the literal values in sync with those that may have been in-lined with old consumer code.

Personal opinion is that it's probably not worth the effort of trying to convert. For one thing, the "public static final int" idiom isn't going away any time soon, given that it's sprinkled liberally all over the JDK. For another, tracking down usages of the original ints is likely to be really unpleasant, given that your classes will compile away the reference so you're likely not to know you've broken anything until it's too late
(by which I mean
class A
{
public static final int MY_CONSTANT=1
}
class B
{
....
i+=A.MY_CONSTANT;
}
gets compiled into
i+=1
So if you rewrite A you may not ever realize that B is broken until you recompile B later.
It's a pretty well known idiom, probably not so terrible to leave it in, certainly better than the alternative.

There is a rumor that the creator of "make" realized that the syntax of Makefiles was bad, but felt that he couldn't change it because he already had 10 users.
Backwards compatibility at all costs, even if it hurts your customers, is a bad thing. SO can't really give you a definitive answer on what to do in your case, but be sure and consider the cost to your users over the long term.
Also think about ways you can refactor the core of your code will keeping the old integer based enums only at the outer layer.

Wait for the next major revision, change everything to enum and provide a script (sed, perl, Java, Groovy, ...) to convert existing source code to use the new syntax.
Obviously this has two drawbacks:
No binary compatibility. How important this one is depends on the use cases, but can be acceptable in the case of a new major release
Users have to do some work. If the work is simple enough, then this too may be acceptable.
In the meantime, add new types as enums and keep old types as ints.

The best would be if you could just fix the published versions, if possible. In my opinion consistency would be the best solution, so you would need to do some refactoring. I personally don't like deprecated things, because they get into way. You might be able to wait until a bigger version release and use those ints until then, and refactor everything in a big project. If that is not possible, you might consider yourself stuck with the ints, unless you create some kinds of wrappers or something.
If nothing helps but you still evolve the code, you end up losing consistency or living with the deprecated versions. In any case, usually at least at some point of time people become fed up with old stuff if it has lost it's consistency and create new from scratch... So you would have the refactoring in the future no matter what.
The customer might scrap the project and buy an other product, if something goes wrong. Usually it is not the customer's problem can you afford refactoring or not, they just buy what is appropriate and usable to them. So in the end it is a tricky problem and care needs to be taken.