How to get javascript object references or referen

问题:

How to get reference count for an object

• Is it possible to determine if a javascript object has multiple references to it?
• Or if it has references besides the one I'm accessing it with?
• Or even just to get the reference count itself?
• Can I find this information from javascript itself, or will I need to keep track of my own reference counters.

Obviously, there must be at least one reference to it for my code access the object. But what I want to know is if there are any other references to it, or if my code is the only place it is accessed. I'd like to be able to delete the object if nothing else is referencing it.

If you know the answer, there is no need to read the rest of this question. Below is just an example to make things more clear.

---

Use Case

In my application, I have a Repository object instance called contacts that contains an array of ALL my contacts. There are also multiple Collection object instances, such as friends collection and a coworkers collection. Each collection contains an array with a different set of items from the contacts Repository.

Sample Code

To make this concept more concrete, consider the code below. Each instance of the Repository object contains a list of all items of a particular type. You might have a repository of Contacts and a separate repository of Events. To keep it simple, you can just get, add, and remove items, and add many via the constructor.

var Repository = function(items) {
  this.items = items || [];
}
Repository.prototype.get = function(id) {
  for (var i=0,len=this.items.length; i<len; i++) {
    if (items[i].id === id) {
      return this.items[i];
    }
  }
}
Repository.prototype.add = function(item) {
  if (toString.call(item) === "[object Array]") {
    this.items.concat(item);
  }
  else {
    this.items.push(item);
  }
}
Repository.prototype.remove = function(id) {
  for (var i=0,len=this.items.length; i<len; i++) {
    if (items[i].id === id) {
      this.removeIndex(i);
    }
  }
}
Repository.prototype.removeIndex = function(index) {
  if (items[index]) {
    if (/* items[i] has more than 1 reference to it */) {
      // Only remove item from repository if nothing else references it
      this.items.splice(index,1);
      return;
    }
  }
}  

Note the line in remove with the comment. I only want to remove the item from my master repository of objects if no other objects have a reference to the item. Here's Collection:

var Collection = function(repo,items) {
  this.repo = repo;
  this.items = items || [];
}
Collection.prototype.remove = function(id) {
  for (var i=0,len=this.items.length; i<len; i++) {
    if (items[i].id === id) {
      // Remove object from this collection
      this.items.splice(i,1);
      // Tell repo to remove it (only if no other references to it)
      repo.removeIndxe(i);
      return;
    }
  }
}

And then this code uses Repository and Collection:

var contactRepo = new Repository([
    {id: 1, name: "Joe"},
    {id: 2, name: "Jane"},
    {id: 3, name: "Tom"},
    {id: 4, name: "Jack"},
    {id: 5, name: "Sue"}
  ]);

var friends = new Collection(
  contactRepo,
  [
    contactRepo.get(2),
    contactRepo.get(4)
  ]
);

var coworkers = new Collection(
  contactRepo,
  [
    contactRepo.get(1),
    contactRepo.get(2),
    contactRepo.get(5)
  ]
);

contactRepo.items; // contains item ids 1, 2, 3, 4, 5 
friends.items;  // contains item ids 2, 4
coworkers.items;  // contains item ids 1, 2, 5

coworkers.remove(2);

contactRepo.items; // contains item ids 1, 2, 3, 4, 5 
friends.items;  // contains item ids 2, 4
coworkers.items;  // contains item ids 1, 5

friends.remove(4);

contactRepo.items; // contains item ids 1, 2, 3, 5 
friends.items;  // contains item ids 2
coworkers.items;  // contains item ids 1, 5

Notice how coworkers.remove(2) didn't remove id 2 from contactRepo? This is because it was still referenced from friends.items. However, friends.remove(4) causes id 4 to be removed from contactRepo, because no other collection is referring to it.

Summary

The above is what I want to do. I'm sure there are ways I can do this by keeping track of my own reference counters and such. But if there is a way to do it using javascript's built-in reference management, I'd like to hear about how to use it.

回答1:

No, no, no, no; and yes, if you really need to count references you will have to do it manually. JS has no interface to this, GC, or weak references.

Whilst you could implement a manual reference-counted object list, it's questionable whether all the extra overhead (in performance terms but more importantly code complexity) is worth it.

In your example code it would seem simpler to forget the Repository, use a plain Array for your lists, and let standard garbage collection take care of dropping unused people. If you needed to get a list of all people in use, you'd just concat the friends and coworkers lists (and sort/uniquify them if you needed to).

回答2:

You may interest to look into reduce functions, and array.map functions. map could be used to help identify where your collections intersect, or if there is an intersection at all. A user defined reduce function could be used like a merge (kinda like overriding the addition operator so that you can apply operation to objects, or merge all collections on "id" if that is how you define your reduce function - then assign the result to your master reference array, I recommend keeping a shadow array that holds all of the root object/values in case you would like to REWIND or something). Note: one must be careful of prototype chains when reducing an object or array. The map function will be very helpful in this case.

I would suggest not to remove the object or record that is in your Repository as you may want to reference it again later. My approach would be to create a ShadowRepository that would reflect all records/objects that have at least one "Reference". From your description and code presented here it appears you are initializing all of the data and storing reference to 1,2,4,5 as appears in your code.

var contactRepo = new Repository([
    {id: 1, name: "Joe"},
    {id: 2, name: "Jane"},
    {id: 3, name: "Tom"},
    {id: 4, name: "Jack"},
    {id: 5, name: "Sue"}
]);
var friends = new Collection(contactRepo,[
    contactRepo.get(2),
    contactRepo.get(4)
]);

var coworkers = new Collection(contactRepo,[
    contactRepo.get(1),
    contactRepo.get(2),
    contactRepo.get(5)
]);

From the initialization of the Repository and the collections, what you are asking "Remove item from repository if there are no references to it" item 3 would need to be removed immediatly. You can however track the references in a few different ways.

I have considered using Object.observe for a similar situation. However, Object.observe does not work in all browsers. I have recently turned to WatchJS

I am working on understanding the code behind Watch.JS to allow a list of observers on an object to be created dynamically this would allow one to also remove an item that is no longer watched, though I suggest to remove the reference at the point of access - What I mean is a variable that shares the immediate lexical scope with an object that has given a single point of reference to it's sibling can be removed making it no longer accessable outside of the object that had exposed the record/item/property/object of it's sibling. With the reference that all of your other references depended on removed access to the underlying data is stopped. I am generating unique id for origin references to avoid accidentally reusing the same one.

Thank you for sharing your question and the structure you are using, it has helped me consider one of my own specific cases where I was generating uniquely identified references to a lexical sibling these unique ids were kept on the ONE object that had scope, After reading here I have reconsidered and decided to expose only one reference then assign that reference to a variable name where ever it is needed such as in creating a watcher or observer or other Collection.