| Re: [cross-project-issues-dev] Performance, 3.8 versus 4.2 |
|
Hi John,
The bugzilla has 85 comments as I write this. It's a pretty convincing essay over the current state of affairs and I think that Andrey Loskutov's last comment raises some interesting questions to be answered by the PMC. Of course I could spend a couple of hours collecting profiler output and add yet another comment but it doesn't seem to be very relevant. A far more viable approach would be to do what Andrey suggests and reinstate the performance regression tests. I'm too surprised that a decision was made to have them discontinued, especially since I just learned that the initial results were quite horrible. The "Close All" and "Close Each" that's been mentioned in this thread both shows a 2000% performance regression for instance. In fact, not counting the tests that are "too fast to be meaningful", there's only one single test that doesn't show a serious regression. Isn't that enough to get some serious improvement work going? I'm back to my original question; why is the 4.2 platform what's being fronted on the Eclipse download page? What can be done about it today? Is the 4.2SR1 anywhere close to be as good as 3.8 (I was initially disappointed when testing the latest milestone)? Can the download page be changed so that 3.8 is more visible? Can we have the 3.8 based EPP packages back? I'm asking because I feel that the current situation is unfair, both to the Eclipse community and to the new 4.2 technology. It should have been incubated longer. As it stands now, it risks to get a very bad reputation. Perhaps resources can be mustered up so that a) tests can be reinstated and b) the regressions can be fixed, in time for Kepler? - thomas On 2012-09-05 15:21, John Arthorne wrote: I suggest anyone having problems to add constructive details on that bug. For example profiler output when repeatedly performing a slow operation, what plugins are installed, whether it is reproducible with vanilla Eclipse SDK, etc. There are some users reporting pervasive slowdowns, and for many others it is performing well. Something like a listener leak could have effects like this in conjunction with particular installed plugins. It takes time after any major release to isolate and resolve problems like this. |