Thanks for bringing this up, Antonio. I'm all for adding such tests.
In addition for identifying performance bottlenecks we need profiling data of the methods in questions. This will ensure that we are optimizing based on facts :) Regards Carsten 2011/10/7 Antonio Sanso <[email protected]>: > Thanks for the feedback Justin, > > > On Oct 7, 2011, at 6:14 PM, Justin Edelson wrote: > >> Antonio- >> I think adding some performance tests is a great idea, although I >> think it'll be challenging to separate Sling performance from the >> underlying JCR performance. >> >> In general, my preference is for microbenchmarks to prove (or >> disprove) a particular design strategy. The changes you describe in >> SLING-2239 are good examples of changes which should be proven out >> with benchmarks before being applied. > > this is exactly what I think and the reports produced can be used to verify > that there isn't any regression in performance while releasing a new bundle > version (with a possible new 'strategy'). > Regards > > Antonio > >> >> Justin >> >> On Fri, Oct 7, 2011 at 10:34 AM, Antonio Sanso <[email protected]> wrote: >>> Hi all, >>> >>> SLING-2239 [0] made me think that it might be useful to have a kind of >>> performance tests as for example >>> the one in Jackrabbit [1] in order to have periodical report as in [2]. >>> >>> Should you like the idea I might try to 'port' [1] to Sling (with all the >>> credits to Jackrabbit obviously :)) >>> >>> WDYT? >>> >>> Regards >>> >>> Antonio >>> >>> >>> [0] https://issues.apache.org/jira/browse/SLING-2239 >>> [1] >>> http://svn.apache.org/repos/asf/jackrabbit/trunk/test/performance/base/src/main/java/org/apache/jackrabbit/performance/ >>> [2] http://people.apache.org/~jukka/jackrabbit/report-2011-08-25/report.html > > -- Carsten Ziegeler [email protected]
