[jira] [Commented] (COLLECTIONS-429) Performance problem in MultiValueMap

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view

[jira] [Commented] (COLLECTIONS-429) Performance problem in MultiValueMap

AD_LB (Jira)

    [ https://issues.apache.org/jira/browse/COLLECTIONS-429?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13566508#comment-13566508 ]

Thomas Neidhart commented on COLLECTIONS-429:

In r1440418 I added a CollectionUtils.containsAll(Collection, Collection) method which provides guaranteed runtime complexity of O(n), regardless of the type of Collection.

Now, this method, similar to the one in the patch, will duplicate the original collection in a HashSet, which may not be desirable in case of really large collections, so I would rather opt to not implement the performance improvement for the referenced Map implementations.

In case someone wants a fast containsAll, he/she should call CollectionUtils.containsAll(), with the downside of more memory usage.

Comments, opinions?

> Performance problem in MultiValueMap
> ------------------------------------
>                 Key: COLLECTIONS-429
>                 URL: https://issues.apache.org/jira/browse/COLLECTIONS-429
>             Project: Commons Collections
>          Issue Type: Bug
>    Affects Versions: 3.2.1
>         Environment: java 1.6.0_24
> Ubuntu 11.10
>            Reporter: Adrian Nistor
>         Attachments: patchFull_AbstractHashedMap.diff, patchFull.diff, patchFull_StaticBucketMap.diff, patchSmall_AbstractHashedMap.diff, patchSmall.diff, patchSmall_StaticBucketMap.diff, Test_AbstractHashedMap.java, TestDifferentParameter.java, Test.java, Test_StaticBucketMap.java
> Hi,
> I am encountering a performance problem in MultiValueMap.  It appears
> in version 3.2.1 and also in revision 1366088.  I attached a test that
> exposes this problem and a patch that fixes it.  On my machine, for
> this test, the patch provides a 1158X speedup.
> To run the test, just do:
> $ java Test
> The output for the un-patched version is:
> Time is 44040
> The output for the patched version is:
> Time is 38
> The attached test shows that, for a "MultiValueMap multi" object, the
> following operation is very slow:
> multi.values().containsAll(toContain);
> "MultiValueMap.values()" returns a "MultiValueMap.Values" object,
> which inherits containsAll() from "java.util.AbstractCollection",
> which has quadratic complexity.
> I attached two patches.  Both patches override containsAll() and
> implement a linear time algorithm.  patchSmall.diff populates a
> HashSet eagerly, and patchFull.diff populates a HashSet lazily.
> patchFull.diff is faster than patchSmall.diff when containsAll()
> returns false after inspecting only a few elements, though in most
> cases they are equally fast.  I am including patchSmall.diff just in
> case you prefer smaller code.
> Note that this problem is different from COLLECTIONS-416.  As
> established in the COLLECTIONS-416 discussion, there the user was
> responsible for using the proper data structures as argument for the
> method.
> For "MultiValueMap.values()", the problem is not related to the
> collection passed as parameter.  The problem will always manifest for
> this method, irrespective of the parameter type.  I attached a test
> (TestDifferentParameter.java) that shows that, even with a HashSet
> parameter, the current problem still manifests (which did not happen
> for COLLECTIONS-416).
> This problem also exists for the two other "Values" classes
> (AbstractHashedMap.Values, StaticBucketMap.Values).  I attached tests
> and patches for these classes as well.  If you want me to file
> separate reports, just let me know.
> Is this truly a performance problem, or am I misunderstanding the
> intended behavior?  If so, can you please confirm that the patches are
> correct?
> Thanks,
> Adrian

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira