2

ConcurrentHashMap documentation says:

Retrieval operations (including get) generally do not block, so may overlap with update operations (including put and remove). Retrievals reflect the results of the most recently completed update operations holding upon their onset. (More formally, an update operation for a given key bears a happens-before relation with any (non-null) retrieval for that key reporting the updated value.)

But I am not able to understand how retrieval operation do not block with update/remove operation for same key ?

"an update operation for a given key bears a happens-before relation with any (non-null) retrieval for that key"

that sort of indicate that concurrent hash map has some way create sequence for read/update/remove operations on same key. But that mean read operations are blocked by update/remove operations.

Not sure what I am missing here.

Improve this question
4
  • May be try some code and clear your doubts; otherwise its just more theory. Do you know what happens-before means (Oracle's Java tutorials has an example and explanation of it)? Commented Jul 7, 2018 at 1:13
  • 1
    Another suggestion: try to think an example application where you might want to use this collection. And, how it can be useful in a particular scenario or solve a problem. Commented Jul 7, 2018 at 2:53
  • Think of an example application where you are using a HashMap and applying sychronization or locks (java.util.concurrent.locks package) to read and update the collection. Perhaps using a ConcurrentHashMap, instead of a HashMap and sychronization, in such case may be a simpler way. Commented Jul 7, 2018 at 3:40
  • Thanks for pointing me to java's tutorial on happen-before. Commented Jul 7, 2018 at 6:18

1 Answer 1

Reset to default
9

But I am not able to understand how retrieval operation do not block with update/remove operation for same key ?

First of all, "how" it does it is an implementation detail. If you really want / need to understand "how", you need to look at the source code. It is available from various places.

But what this is actually saying is that:

  • get doesn't block, and
  • what you see when you do a get is the value from the most recently completed put operation for that key.

So if you do a get while a put is in progress, you may see the value from the previous put.

Note that the 2nd bullet point explains (implicitly) why it is possible for get to be non-blocking. There is no need for get(42) to wait for a put(42, value) that is currently in progress to finish ... since the get call is allowed to return the previous value.

TL;DR - there is no need to block.

The stuff about "happens before" relationships is relating this to the semantics of the Java Memory Model. This may be important if you are doing a deep analysis of your code. But for a shallow understanding of what ConcurrentHashMap does, you can ignore it.

... that sort of indicate that concurrent hash map has some way create sequence for read/update/remove operations on same key.

It doesn't imply that at all. But to understand what the statement you quoted really means, you need a need a good understanding of the Java Memory Model. And I don't think it is feasible to convey that to you in a StackOverflow Q&A. I recommend you take a few hours to read and understand the JMM ... by reading the JLS. Then, do it again.

This is NOT stuff where an intuitive understanding is sufficient.

Improve this answer
Sign up to request clarification or add additional context in comments.

1 Comment

Thanks. I actually wanted to implement some sort of custom concurrent trie and wanted to know how ConcurrentHashMap keep read unblocked. Looks like might not be so simple as I assumed :)

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.