Re: NodeBuilder#setProperty using the same value
On 5.6.13 13:43, Jukka Zitting wrote: Hi, On Wed, Jun 5, 2013 at 3:32 PM, Alex Parvulescu wrote: I'm wondering if we could turn the NodeBuilder#setProperty into a noop when setting the same value on a property. I don't see any bigger issues with this, though it would be nice to have a benchmark that verifies that the added complexity (though small in this case) is worth it. Upfront equality check on all such calls might end up being more expensive then allowing a property to be set to its own value once in a while. Michael
Re: Some more benchmarks
Hi, On Fri, May 31, 2013 at 3:14 PM, Jukka Zitting wrote: > It looks like we have a performance regression in ReadPropertyTest. > Quick profiling shows a lot of the time seems to be going to > MemoryNodeBuilder$ConnectedHead.update(), which is weird since we're > only reading and thus the related MNB head should be unconnected. I'll > investigate. Revision 1490258 fixed the issue. The updated results are: Apache Jackrabbit Oak 0.9-SNAPSHOT # ReadPropertyTest min 10% 50% 90% max N Jackrabbit40 41 41 42 971448 Oak-Default 11 12 12 14 194804 Oak-Mongo 17 18 18 20 1233128 Oak-Segment 94 94 96 98 136 622 Oak-Tar 11 11 12 13 175121 BR, Jukka Zitting
Re: svn commit: r1490258 - /jackrabbit/oak/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/plugins/memory/MemoryNodeBuilder.java
On 6.6.13 13:41, ju...@apache.org wrote:
Author: jukka
Date: Thu Jun 6 12:41:28 2013
New Revision: 1490258
URL: http://svn.apache.org/r1490258
Log:
OAK-781: Clarify / fix effects of MISSING_NODE as base state of NodeBuilder
Fix the performance issue of a ConnectedHead never actually updating
the head state after a rebase. It would just repeatedly create new
UnconnectedHead instances without ever setting the head field. This
commit makes updating the head field the responsibility of the head()
method, which by centralizing the state transition should help avoid
potential other similar issues.
Thanks for finding this. Looks better now.
Modified:
jackrabbit/oak/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/plugins/memory/MemoryNodeBuilder.java
URL:
http://svn.apache.org/viewvc/jackrabbit/oak/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/plugins/memory/MemoryNodeBuilder.java?rev=1490258&r1=1490257&r2=1490258&view=diff
==
---
jackrabbit/oak/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/plugins/memory/MemoryNodeBuilder.java
(original)
+++
jackrabbit/oak/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/plugins/memory/MemoryNodeBuilder.java
Thu Jun 6 12:41:28 2013
@@ -142,7 +142,11 @@ public class MemoryNodeBuilder implement
* @return head of this builder
*/
private Head head() {
-return head.update();
+Head newHead = head.update();
+if (newHead != head) {
+head = newHead;
+}
+return newHead;
}
Isn't this the same as
return head = head.update()
??
Michael
Re: svn commit: r1490258 - /jackrabbit/oak/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/plugins/memory/MemoryNodeBuilder.java
Hi,
On Thu, Jun 6, 2013 at 4:01 PM, Michael Dürig wrote:
> On 6.6.13 13:41, ju...@apache.org wrote:
>> private Head head() {
>> -return head.update();
>> +Head newHead = head.update();
>> +if (newHead != head) {
>> +head = newHead;
>> +}
>> +return newHead;
>> }
>
> Isn't this the same as
>
> return head = head.update()
It is, except for the nice side-effect of being able to easily set a
breakpoint on state transitions.
BR,
Jukka Zitting
jackrabbit-oak build #1700: Broken
Build Update for apache/jackrabbit-oak - Build: #1700 Status: Broken Duration: 1150 seconds Commit: fa27d3c33ea57a8f3b34bceb12937b8b6a28ce3b (trunk) Author: Thomas Mueller Message: OAK-13 Use a daemon thread to avoid blocking the build. git-svn-id: https://svn.apache.org/repos/asf/jackrabbit/oak/trunk@1490266 13f79535-47bb-0310-9956-ffa450edef68 View the changeset: https://github.com/apache/jackrabbit-oak/compare/09d79d819670...fa27d3c33ea5 View the full build log and details: https://travis-ci.org/apache/jackrabbit-oak/builds/7840285 -- sent by Jukka's Travis notification gateway
jackrabbit-oak build #1702: Broken
Build Update for apache/jackrabbit-oak - Build: #1702 Status: Broken Duration: 669 seconds Commit: fb96bd6b92e3f3fff91bc9fb92cc5987d2715e96 (trunk) Author: Jukka Zitting Message: OAK-641: Improved benchmark tooling Better formatting of hprof analysis reports git-svn-id: https://svn.apache.org/repos/asf/jackrabbit/oak/trunk@1490352 13f79535-47bb-0310-9956-ffa450edef68 View the changeset: https://github.com/apache/jackrabbit-oak/compare/e5c15e87e81a...fb96bd6b92e3 View the full build log and details: https://travis-ci.org/apache/jackrabbit-oak/builds/7847527 -- sent by Jukka's Travis notification gateway
jackrabbit-oak build #1703: Fixed
Build Update for apache/jackrabbit-oak - Build: #1703 Status: Fixed Duration: 1602 seconds Commit: 924baad7277f7fc75c699d98f4d17f0586655cfc (trunk) Author: Jukka Zitting Message: OAK-799: Fail-fast for content diffs Streamline EditorDiff to reduce the number of stack frames in profiler reports git-svn-id: https://svn.apache.org/repos/asf/jackrabbit/oak/trunk@1490361 13f79535-47bb-0310-9956-ffa450edef68 View the changeset: https://github.com/apache/jackrabbit-oak/compare/fb96bd6b92e3...924baad7277f View the full build log and details: https://travis-ci.org/apache/jackrabbit-oak/builds/7848103 -- sent by Jukka's Travis notification gateway
