[ https://jira.jboss.org/jira/browse/JBSEAM-4244?page=com.atlassian.jira.pl... ] Riccardo Serafin commented on JBSEAM-4244: ------------------------------------------ From: http://www.seamframework.org/Community/212PerformanceRegressionInMessages The problem is the following: I've been profiling my application and since I switched to 2.1.2 I've noticed a huge amount of calls to Messages.AbstractMap.entrySet() and the related performance problem, as this method creates the entry set recurring all entries of all entry sets of the underlying properties maps. As a result, I get this method as the first one in the profiler in terms of consumed time. After looking at the code, I noticed that: 1) the implementation of containsKey that was included in 2.1.1 has been removed. This implementation used the faster get() method instead of the slower entrySet() method, which is used by the default AbstractMap implementation, to determine if the key is present. 2) The method resolveInMap in the SeamELResolver is now using a call to containsKey before trying to access the value. So, to test if my theory was correct, I've replaced the Messages class with my own, which extends the default one and adds the containsKey implementation copied from 2.1.1. Profiling results has gone back to "normal". To test, I guess you can just try to profile the execution of a containsKey on the current implementation, then put back the previous containsKey override and profile again. You should be noticing the gain. Related issues: JBSEAM-2192 - This is the one where the containsKey fix was introduced. JBSEAM-3916 - This is the one where the call to containsKey as been introduced in SeamELResolver (which worsen the problems). -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://jira.jboss.org/jira/browse/JBSEAM-4244?page=com.atlassian.jira.pl... ] Riccardo Serafin commented on JBSEAM-4244: ------------------------------------------ Thanks, i've tested the patch and it looks like this solves the problem. At least as far my test page was involved. Just to give you the idea, the total time spent in SeamELResolver.resolveInMap method when reloading my page has gone from 0.2ms to 1300ms from 2.1.1 to 2.1.2 and back to 0.2ms with 2.2.0-SNAPSHOT. I've attached an excel file with the results of my test. One problem that is still there, however, is that the containsKey over the messages does still suffer from the performance problem. So i believe the author of JBSEAM-2192 will still be quite unhappy. Moreover, I wanted to ask why you had to change the AbstractMap.get method, and why in the current version you interpolate the obtained from the bundle, given that SeamResouceBundle is already interpolating. Finally, as it seems you are dealing with the SeamELResolver a lot, I was wondering if have you had time to look at this other post of mines related to resolver performance (http://www.seamframework.org/Community/SeamELResolverPerformanceIssues)? Thanks a lot in the meantime. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ https://jira.jboss.org/jira/browse/JBSEAM-4244?page=com.atlassian.jira.pl... ] Norman Richards closed JBSEAM-4244. ----------------------------------- Resolution: Done -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira