You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@commons.apache.org by "Adrian Nistor (JIRA)" <ji...@apache.org> on 2012/06/19 22:38:44 UTC

[jira] [Created] (COLLECTIONS-410) SetUniqueList.addAll() is very slow

Adrian Nistor created COLLECTIONS-410:
-----------------------------------------

             Summary: SetUniqueList.addAll() is very slow
                 Key: COLLECTIONS-410
                 URL: https://issues.apache.org/jira/browse/COLLECTIONS-410
             Project: Commons Collections
          Issue Type: Bug
         Environment: java 1.6.0_24
Ubuntu 11.10
            Reporter: Adrian Nistor
         Attachments: Test.java, patch.diff

Hi,

I am encountering a performance problem in SetUniqueList.addAll().  It
appears in revision 1351837 (19 June 2012).  I attached a test that
exposes this problem and a patch that fixes it.  On my machine, for
this test, the patch provides a 540X speedup.

To run the test, just do:

$ java Test

The output for the un-patched version is:
Time is 2706

The output for the patched version is:
Time is 5

As the patch shows, the problem is that
SetUniqueList.addAll(int index, Collection<? extends E> coll)
performs:
"add(index, e)" for each element in "coll". This is very expensive, 
because each "add(index, e)" performs a 
LinkedList.add(int index, E element), which requires traversing the 
LinkedList to find the index.

The patched version avoids this cost by inserting all the elements at
once, thus performing only one insert.

Is this a bug? If so, can you please confirm that the patch is
correct?

Thanks,

Adrian


--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

[jira] [Commented] (COLLECTIONS-410) SetUniqueList.addAll() is very slow

Posted by "Hudson (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/COLLECTIONS-410?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13397744#comment-13397744 ] 

Hudson commented on COLLECTIONS-410:
------------------------------------

Integrated in commons-collections #27 (See [https://builds.apache.org/job/commons-collections/27/])
    [COLLECTIONS-410] Improved performance of SetUniqueList.addAll(index, coll). Thanks to Adrian Nistor for reporting and providing a patch. (Revision 1352243)

     Result = UNSTABLE
tn : http://svn.apache.org/viewvc/?view=rev&rev=1352243
Files : 
* /commons/proper/collections/trunk/src/main/java/org/apache/commons/collections/list/SetUniqueList.java

                
> SetUniqueList.addAll() is very slow
> -----------------------------------
>
>                 Key: COLLECTIONS-410
>                 URL: https://issues.apache.org/jira/browse/COLLECTIONS-410
>             Project: Commons Collections
>          Issue Type: Bug
>         Environment: java 1.6.0_24
> Ubuntu 11.10
>            Reporter: Adrian Nistor
>             Fix For: 4.0
>
>         Attachments: Test.java, patch.diff
>
>
> Hi,
> I am encountering a performance problem in SetUniqueList.addAll().  It
> appears in revision 1351837 (19 June 2012).  I attached a test that
> exposes this problem and a patch that fixes it.  On my machine, for
> this test, the patch provides a 540X speedup.
> To run the test, just do:
> $ java Test
> The output for the un-patched version is:
> Time is 2706
> The output for the patched version is:
> Time is 5
> As the patch shows, the problem is that
> SetUniqueList.addAll(int index, Collection<? extends E> coll)
> performs:
> "add(index, e)" for each element in "coll". This is very expensive, 
> because each "add(index, e)" performs a 
> LinkedList.add(int index, E element), which requires traversing the 
> LinkedList to find the index.
> The patched version avoids this cost by inserting all the elements at
> once, thus performing only one insert.
> Is this a bug? If so, can you please confirm that the patch is
> correct?
> Thanks,
> Adrian

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

[jira] [Updated] (COLLECTIONS-410) SetUniqueList.addAll() is very slow

Posted by "Adrian Nistor (JIRA)" <ji...@apache.org>.
     [ https://issues.apache.org/jira/browse/COLLECTIONS-410?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Adrian Nistor updated COLLECTIONS-410:
--------------------------------------

    Attachment: Test.java
                patch.diff
    
> SetUniqueList.addAll() is very slow
> -----------------------------------
>
>                 Key: COLLECTIONS-410
>                 URL: https://issues.apache.org/jira/browse/COLLECTIONS-410
>             Project: Commons Collections
>          Issue Type: Bug
>         Environment: java 1.6.0_24
> Ubuntu 11.10
>            Reporter: Adrian Nistor
>         Attachments: Test.java, patch.diff
>
>
> Hi,
> I am encountering a performance problem in SetUniqueList.addAll().  It
> appears in revision 1351837 (19 June 2012).  I attached a test that
> exposes this problem and a patch that fixes it.  On my machine, for
> this test, the patch provides a 540X speedup.
> To run the test, just do:
> $ java Test
> The output for the un-patched version is:
> Time is 2706
> The output for the patched version is:
> Time is 5
> As the patch shows, the problem is that
> SetUniqueList.addAll(int index, Collection<? extends E> coll)
> performs:
> "add(index, e)" for each element in "coll". This is very expensive, 
> because each "add(index, e)" performs a 
> LinkedList.add(int index, E element), which requires traversing the 
> LinkedList to find the index.
> The patched version avoids this cost by inserting all the elements at
> once, thus performing only one insert.
> Is this a bug? If so, can you please confirm that the patch is
> correct?
> Thanks,
> Adrian

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

[jira] [Commented] (COLLECTIONS-410) SetUniqueList.addAll() is very slow

Posted by "Thomas Neidhart (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/COLLECTIONS-410?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13397860#comment-13397860 ] 

Thomas Neidhart commented on COLLECTIONS-410:
---------------------------------------------

The failing unit test on hudson/jenkins has been fixed in r1352264.

                
> SetUniqueList.addAll() is very slow
> -----------------------------------
>
>                 Key: COLLECTIONS-410
>                 URL: https://issues.apache.org/jira/browse/COLLECTIONS-410
>             Project: Commons Collections
>          Issue Type: Bug
>         Environment: java 1.6.0_24
> Ubuntu 11.10
>            Reporter: Adrian Nistor
>             Fix For: 4.0
>
>         Attachments: Test.java, patch.diff
>
>
> Hi,
> I am encountering a performance problem in SetUniqueList.addAll().  It
> appears in revision 1351837 (19 June 2012).  I attached a test that
> exposes this problem and a patch that fixes it.  On my machine, for
> this test, the patch provides a 540X speedup.
> To run the test, just do:
> $ java Test
> The output for the un-patched version is:
> Time is 2706
> The output for the patched version is:
> Time is 5
> As the patch shows, the problem is that
> SetUniqueList.addAll(int index, Collection<? extends E> coll)
> performs:
> "add(index, e)" for each element in "coll". This is very expensive, 
> because each "add(index, e)" performs a 
> LinkedList.add(int index, E element), which requires traversing the 
> LinkedList to find the index.
> The patched version avoids this cost by inserting all the elements at
> once, thus performing only one insert.
> Is this a bug? If so, can you please confirm that the patch is
> correct?
> Thanks,
> Adrian

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

[jira] [Resolved] (COLLECTIONS-410) SetUniqueList.addAll() is very slow

Posted by "Thomas Neidhart (JIRA)" <ji...@apache.org>.
     [ https://issues.apache.org/jira/browse/COLLECTIONS-410?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Thomas Neidhart resolved COLLECTIONS-410.
-----------------------------------------

       Resolution: Fixed
    Fix Version/s: 4.0

Applied patch in r1352243.

Thanks for reporting!
                
> SetUniqueList.addAll() is very slow
> -----------------------------------
>
>                 Key: COLLECTIONS-410
>                 URL: https://issues.apache.org/jira/browse/COLLECTIONS-410
>             Project: Commons Collections
>          Issue Type: Bug
>         Environment: java 1.6.0_24
> Ubuntu 11.10
>            Reporter: Adrian Nistor
>             Fix For: 4.0
>
>         Attachments: Test.java, patch.diff
>
>
> Hi,
> I am encountering a performance problem in SetUniqueList.addAll().  It
> appears in revision 1351837 (19 June 2012).  I attached a test that
> exposes this problem and a patch that fixes it.  On my machine, for
> this test, the patch provides a 540X speedup.
> To run the test, just do:
> $ java Test
> The output for the un-patched version is:
> Time is 2706
> The output for the patched version is:
> Time is 5
> As the patch shows, the problem is that
> SetUniqueList.addAll(int index, Collection<? extends E> coll)
> performs:
> "add(index, e)" for each element in "coll". This is very expensive, 
> because each "add(index, e)" performs a 
> LinkedList.add(int index, E element), which requires traversing the 
> LinkedList to find the index.
> The patched version avoids this cost by inserting all the elements at
> once, thus performing only one insert.
> Is this a bug? If so, can you please confirm that the patch is
> correct?
> Thanks,
> Adrian

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira