I try to figure out how the WeakHashMap cleans up after garbare collection.
As many of you may know, the WeakHashMap entry is removed automatically when its key becomes garbage collected. But, for instance, if I do something like this:
List<WeakReference<Main>> list = new ArrayList<>();
list.add(new WeakReference<>(new Main()));
System.gc();
Thread.sleep(1000);
list.get(0).get(); //null - WeakReference referent has been removed
list.get(0); //empty WeakReference object is still present in the List
ArrayList doesn't clean the empty WeakReference objects, but why WeakHashMap does? Which component is responsible for this automatic entry removing. I don't see any code in the WeakHashMap sources which could do that.
WeakHashMap cleans itself up. Calls to getTable() (itself called by get, put, etc.), size() and resize(int) all first call expungeStaleEntries, which iterates a ReferenceQueue of GC'd Entries and removes them from the table. The garbage collector is responsible for putting unreachable WeakReference objects on the queue.
Related
I have a weakly-referenced array of weak references to further objects like so:
public class Foo{
WeakReference<WeakReference<Bar>[]> cache;
}
public class Bar{
private final WeakReference<Bar>[] ownerCache;
}
The array itself is weakly referred for reasons I will not go into right now. I want to make sure it is not garbage collected before ANY of the Bar objects reachable from it. In other words, it must exist in memory as long as any Bar object exists that can be reached from it. Then, if no Bar objects exist any more, I am better off if the array is also garbage collected. (Bar objects may or may not be strongly reachable from elsewhere.) I did this by referring to the cache array in a field inside all Bar objects. If the field is sufficient in making the array strongly reachable, it is not garbage collected. However, my code never actually uses that field and I can not make it public. (I get the "unused" warning on it.) I am afraid that the existence of such field is terminated either during compile time or run time, or it could get special treatment from the garbage collector that I am unaware of.
Is this the right solution? Does this solution achieve what I want regardless of the garbage collector or JVM implementation? If not, what would be a better method?
Here are a couple of ideas.
If you control the Bar class, and each instance is referenced by no more than one array, you could add a reference from a Bar instance to the array. Reachable Bar instances will prevent the array from being collected.
Alternatively, you could:
Construct a reference queue for weak references to Bar instances.
ReferenceQueue<Bar> m_refQueue = new ReferenceQueue<>();
Construct each WeakReference with a reference to that queue.
new WeakReference<Bar>( myBar, m_refQueue );
Periodically poll that queue for available collectable instances, and remove them from your collection.
You could make the collection itself a resizable data structure, avoiding the need to collect it.
public class Foo {
final #Nonnull List<WeakReference<Bar>> cache = new ArrayList<>();
// Or you could use an IdentityHashSet from a third-party library.
}
EDIT
As suggested by #Holger below, if an ordered list of references is not needed, your collection of WeakReference can be a java.util.WeakHashMap, used as a set. The keys are weak references; the values can be null. The map is a resizable data structure, so you can simply hold an ordinary reference to the map.
public class Foo {
final #Nonnull WeakHashMap<WeakReference<Bar>,Object> cache
= new WeakHashMap<>();
There is a WeakHashMap instance initialized with, for example, 500 entries. Now, its keys have not been referenced anywhere in the running application for a day or so. Will this map's entries be removed automatically after a certain time gets passed?
My understanding is that if key is not referenced then corresponding entries will be removed from the map.
It will be removed when GC runs if your key is not referenced anywhere: (Reference)
Hash table based implementation of the Map interface, with weak keys.
An entry in a WeakHashMap will automatically be removed when its key
is no longer in ordinary use. More precisely, the presence of a
mapping for a given key will not prevent the key from being discarded
by the garbage collector, that is, made finalizable, finalized, and
then reclaimed. When a key has been discarded its entry is effectively
removed from the map, so this class behaves somewhat differently from
other Map implementations.
The removal time is unknown:
Each key object in a WeakHashMap is stored indirectly as the referent
of a weak reference. Therefore a key will automatically be removed
only after the weak references to it, both inside and outside of the
map, have been cleared by the garbage collector.
But be careful, some object like boxed Integer of small integers like -127-> 127 are cached by JVM so if you use autoboxed Integer key, it will never be removed from the Map.
Well, first we narrow down the question.
QUESTION: We have a WeakHashMap in which we have some entries. Will those entries will be garbage collected if the entries are not being used?
Ref code:
WeakHashMap<Object, Object> wkMap = new WeakHashMap<>()
Object obj1 = new Object();
Object obj2 = new Object();
Objcet obj1Meta = new Object();
Objcet obj2Meta = new Object();
wkMap.put(obj1, obj1Meta);
wkMap.put(obj2, obj2Meta);
First of all, it's not about being used, neither it has any relation with time: it's about whether the reference to the map (wkMap in this case) is in scope; if not, then the entire map is eligible for garbage collection, that's quite straightforward. But if not, then...
One thing we need to check is whether the objects which are weakly referenced by the keys of the map are already garbage collected or not. In this case obj1 and obj2.
If these objects have not been garbage collected, then their corresponding entries will be there in the map. Garbage collector is not going to reclaim. Again straightforward.
Now the tricky case: the objects referenced weakly by obj1, obj2 have been garbage collected. There is no need of their metadata present in the map wkMap. Ideally they should be garbage collected, and eventually they are. But the question is how...
Step by Step
The objects referenced weakly by obj1, obj2 become eligible for garbage collection
The garbage collector collects the objects; at this point, the garbage collector checks whether there are any weak references to the object it's collecting. In our case we have two: keys of two entries in the weak hash map wkMap.
If GC finds some weak references to the object it's collecting, it then checks whether those references have any ReferenceQueue attached to it. If there is any then GC puts the weak reference to that ReferenceQueue. GC is done.
Until now, the entries are there in the map and they are not eligible for garbage collection. And it will be there in the map until someone manually makes the keys set to null. Wait, then who does that? Let's see next:
That manual clean-up is done by WeakHashMap itself. Let us check the size() code inside WeakHashMap:
public int size() {
if (size == 0)
return 0;
expungeStaleEntries();
return size;
}
Concentrate on expungeStaleEntries(); it is the one which removes all the entries from the map which are there in the ReferenceQueue as well, and the entries become eligible for garbage collection (a single reference queue is attached to all the weak references used as a key in the map). Check expungeStaleEntries() code as well.
Now in a nutshell, if from your code you call some method on the WeakHashMap, which internally calls this expungeStaleEntries() method, only then will the entries become eligible for garbage collection.
List of methods which call expungeStaleEntries()
size()
reSize()
isEmpty()
putAll()
etc...
Hope this makes things clear.
Will this map's entries be removed automatically after a certain time gets passed?
It depends on when the Garbage Collector comes. There is no guarantee that it reclaims "garbage" even once a day.
The behavior of the WeakHashMap class depends in part upon the actions of the garbage collector, so several familiar (though not required) Map invariants do not hold for this class. Because the garbage collector may discard keys at any time, a WeakHashMap may behave as though an unknown thread is silently removing entries.
JDK 10 - WeakHashMap
In this code
WeakHashMap<Group,String> map = new WeakHashMap<>();
Group group1 = new Group();
Group group2 = new Group();
map.put(group1,"one");
map.put(group2,"two");
System.out.println(map);
group1 = null;
System.gc();
System.out.println(map);
In the first print statement , you will see that hashmap has two elements and in the second print statement it will have only one element. Because the reference to the first element is now null. So yes all keys whose references point to null will get removed the next time GC runs.
{Group#53e25b76=one, Group#73a8dfcc=two}//First print
{Group#73a8dfcc=two}//Second print
I hava a test application with one class that stores a map and a button that invokes method of that class:
Map<Object, Object> weakMap = new WeakHashMap<Object, Object>();
The button does this:
public void fillWeakHashMap(int size) {
List<String> createObjects = (List<String>) createObject(size);
for (Object ob : createObjects) {
weakMap.put(ob, ob);
}
}
So the main class with the map lives but objects which I put to it are created in separate method and since it is a WeakHashMap I thought that after exit from the method fillWeakHashMap the keys / objects in map would be cleared by the GC.
I opened the VisualVM->VisualGC and clicked Monitor->PerformGC 10 times and the Old space is almost full meaning that GC isnt clearing those objects. So what am I doing wrong?
The first sentence of WeakHashMap's javadoc says:
Hash table based implementation of the Map interface, with weak keys. An entry in a WeakHashMap will automatically be removed when its key is no longer in ordinary use. More precisely, the presence of a mapping for a given key will not prevent the key from being discarded by the garbage collector, that is, made finalizable, finalized, and then reclaimed. When a key has been discarded its entry is effectively removed from the map, so this class behaves somewhat differently from other Map implementations.
and somewhat further down, it writes:
The value objects in a WeakHashMap are held by ordinary strong references. Thus care should be taken to ensure that value objects do not strongly refer to their own keys, either directly or indirectly, since that will prevent the keys from being discarded.
That is, only the keys are weakly referenced, but the values are strongly referenced. In your code, each key is also used a value, therefore strongly referenced, and therefore not garbage collected.
You are keeping a reference for each object in the map like ob is still referring itself that is why nothing is being cleared, since GC cant claimed that those objects are not is use and can be cleaned.
What is the need arises for introducing Weak HashMap when there is already other implementations available.
In short i have two issues :
Why jdk has WeakHashMap when there is HashMap and Concurrent HashMap in java ?
What is the use of it in real life applications ?
EDIT :
Though WeakHashmap key is a weak references but still they refer to
something than on what basis GC discard keys in WeakHashMap.
One Common use of WeakReferences and WeakHashMaps in particular is for
adding properties to objects. Occasionally you want to add some
functionality or data to an object but subclassing and/or composition
are not an option in that case the obvious thing to do would be to
create a hashmap linking the object you want to extend to the property
you want to add. then whenever you need the property you can just look
it up in the map. However, if the objects you are adding properties to
tend to get destroyed and created a lot, you can end up with a lot of
old objects in your map taking up a lot of memory
If you use a WeakHashMap instead the objects will leave your map as
soon as they are no longer used by the rest of your program, which is
the desired behavior.
The key of a WeakHashMap has weak reference. If the key has been garbage collected, then the entry in WeakHashMap object will automatically be deleted. It does not happen in normal HashMap. The entry will not be deleted if the key is garbage collected.
In the example I have taken one HashMap and one WeakHashMap. I will put entry in both the object and then later we will make the reference key as null and then garbage collected. And again check the entry. In the HashMap object entry will be there but in WeakHashMap object there will not be entry present.
import java.util.HashMap;
import java.util.Map;
import java.util.WeakHashMap;
public class WeakHashMapTest {
public static void main(String[] args) {
Map hashMap= new HashMap();
Map weakHashMap = new WeakHashMap();
String keyHashMap = new String("keyHashMap");
String keyWeakHashMap = new String("keyWeakHashMap");
hashMap.put(keyHashMap, "Ankita");
weakHashMap.put(keyWeakHashMap, "Atul");
System.gc();
System.out.println("Before: hash map value:"+hashMap.get("keyHashMap")+" and weak hash map value:"+weakHashMap.get("keyWeakHashMap"));
keyHashMap = null;
keyWeakHashMap = null;
System.gc();
System.out.println("After: hash map value:"+hashMap.get("keyHashMap")+" and weak hash map value:"+weakHashMap.get("keyWeakHashMap"));
}
}
The output would be:
Before: hash map value:Ankita and weak hash map value:Atul
After: hash map value:Ankita and weak hash map value:null
More info:
When would you use a WeakHashMap or a WeakReference?
WeakHashMap vs. HashMap
unlike the normal collections, which hold data until you decide to clear it out, data in WeakHashMap may be removed at any point in time and without notice when the JVM decides it wants the memory back. this makes it suitable for all sorts of caching purposes.
you can read up on weak references (and various other related types) here
First you should understand the purpose of WeakReferences. Once you understand that then reading the WeakedHashMap definition from docs clearly tells its purpose
A hashtable-based Map implementation with weak keys. An entry in a
WeakHashMap will automatically be removed when its key is no longer in
ordinary use. More precisely, the presence of a mapping for a given
key will not prevent the key from being discarded by the garbage
collector, that is, made finalizable, finalized, and then reclaimed.
When a key has been discarded its entry is effectively removed from
the map, so this class behaves somewhat differently from other Map
implementations.
You can use an object as a key without preventing its collection.
From the WeakHashMap documentation:
When a key has been discarded its entry is effectively removed from the map, so this class behaves somewhat differently from other Map implementations.
They provide a a way of using an object as a key without creating a strong reference to it. This is good practice where you don't want to hinder JVM's ability to garbage collect the object but yet still want the ability to track some aspect of the object. This makes them ideal for caching or storing metadata about the object.
From the answers, the only cache feature is automatic eviction on GC and it probably won't happen during young gen GC but in tenured old gen GC for the usecase of caching large files that are used for sometime. Is the entry evicted on any GC or only when tenured generation is full?
Besides unpredictable eviction there is no size constraint that can be specified on the map, so any cache library with pre-defined cache size (number of elements and memory size) along with an eviction strategy seems like a better option to me as most caches also include a provision to reduce size/flush more elements when heap usage goes beyond a certain size. These also provide consistency during concurrency.
So, I wonder if there is a good use case to use this besides the risk of memory leaks and missed hits due to bad design.
Below is the code
ArrayList arList = someMethod();// returning ArrayList with customDO objects
Now somewhere in different class I am getting data from this arList
CustomDo custDO= (CustomDO)arList.get(0);
Will the arList be alive as long as custDO is alive ? If yes, will below piece of code help
CustomDO custDO = ((CustomDO)arList.get(0)).cloneMe();
// where cloneMe has defintion as return ((CustomDO)super.clone());
// CustomDo implements Cloneable
Is there a better way to keep a copy of only the first element of arList and discard the list so that it can be collected by garbage collector ?
Is there a better to keep a copy of only the first element of arList and discard the list so that it can be collected by garbage collector ?
You don't have to make a copy of the list element. As long as you have another reference to it, it will not be garbage-collected, even if the list you got it from is. And the list will be garbage-collected as soon as you remove all references to it.
There is no need in Java to clone anything just to make sure that the object does not disappear. In Java a reference to an object is always valid. It cannot happen that the data for a live reference gets invalid.
You only want to make a copy (clone) if you are afraid that other people who reference the same object might change its contents (calling some setter on it) in ways that would cause trouble for you (or you want to have a private copy to change it without affecting others).
// reference to first object
CustomDO custDO = ((CustomDO)arList.get(0));
// let arList be garbage collected
arList = null;
Another thing you should know is that Collections clone() methods do a shallow (flat) copy. Sometimes you need to have deep copies (to allow modifing them independedly)
As long as you have access to CustomDO custDO object, it will not be garbage collected. The list can be garbage collected if there is no reference to it.
The ArrayList is an ordinary Object, and only references to this object will keep the list alive. Of course, as long as the list is alive, all its elements are also alive, but the converse does not hold.