Community
Participate
Working Groups
There is about 40-70% performance regression in the org.eclipse.ant.tests.ui.editor.performance.SeparateVMTests and around 10% regression in OpenAntEditor. This is first seen on 12th Feb.
The regression began once the update to Ant 1.8.2 went live. I was expecting a bit of a dip in the editor performance as we changed the caching scheme to not hold onto a huge DOM object, and in doing so we incur a slight bit more processing which accounts for the few extra milliseconds extra. I am still looking into the separate VM regression. We made some changes to do extra work cleaning up after and during launches to avoid the incremental OOME problem we have been plagued with (bug 130545), but I have no hard evidence yet was is causing such a large regression.
Ping.
No changes are planned here for 3.7, moving to 3.8 for now.
no changes planned here for 3.8 either. We can compare the performance change when we update to Ant 1.8.3 as well.
Bug is getting to too, we have to look again once the performance test are running again.