This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.
Please look at performance test results on Ubuntu ( http://nyx-intel3:8080/hudson/job/cnd-perf-ubuntu-x86/lastBuild/performance ) and notice performance regressions almost in all tests since build #365 (after switching to 7.0.1)
Please evaluate
Please provide steps to reproduce bug.
It seems that the main problem is ProjectViewAfterParsing test. Steps to reproduce: - open a project (loki, LiteSQL, MySQL, Python etc) - wait until the project is parsed - expand some specific source node in projects view Result: Previously in NB 6.9.1 and NB 7.0 it took < 0.5 sec. and now it takes > 2.5 sec.
(In reply to comment #3) > Result: Previously in NB 6.9.1 and NB 7.0 it took < 0.5 sec. and now it takes > > 2.5 sec. Definitely it is not P2.
Paths to folders for test projects: clucene.folder=clucene-0.9.10/src/CLucene/search litesql.folder=litesql-0.3.1/src/library boost.folder=boost_1_36_0/boost loki.folder=loki-0.1.5/src mysql.folder=mysql-5.0.18/sql perl.folder=perl-5.8.5 python.folder=Python-2.4/Modules
Integrated into 'main-golden', will be available in build *201105260401* on http://bits.netbeans.org/dev/nightly/ (upload may still be in progress) Changeset: http://hg.netbeans.org/main/rev/58927df6de62 User: Alexander Simon <alexvsimon@netbeans.org> Log: fixing Bug #198814 - Performance regressions in 7.0.1 (transplanted from d13432f323bc40f96fc77215af2d8374e6a7b764)
It seems benchmark ProjectViewAfterParsing is wrong. It measures random noise with 2 seconds. For example results for boost: NB6.9 283 NB7.0 387 b365 414 b366 2267 b367 263 b370 292 b371 336 b372 2177 So you can see that builds b366 and b372 contain a 2 second noise. Please fix your benchmark ProjectViewAfterParsing. Or compare separately measurements with noise and without noise. By the way the noise is not reproducible manually.
Please be careful with your definitions. It's not a noise as the overall performance index for ProjectViewAfterParsing shows the regression. It's probably not a regression in CND part of NB but something is definitely prevents ProjectsView from working faster. BTW MySQL gives: NB 6.9.1 361 nb 7.0 763 #365 2806 #366 2893 #367 2740 #370 2725 #371 2673 #372 2921 #373 2702 #375 2644 #376 2656
issue is not reproducible manually.
Created attachment 108560 [details] profiler log