From: Manfred U. <uss...@ic...> - 2006-01-27 11:49:56
|
Slava, Slava Pestov <sl...@je...> schrieb am Thu, 26 Jan 2006 20:46:24 -0500: > In jEdit 4.3pre3 I introduced some new display code which performs > custom double buffering instead of using Swing double buffering. The > gain is that scrolling can be done via a bitblit, instead of > repainting lines of text. While on Mac OS X and X11 this resulted in > speed gains due to poor AWT text rendering performance there, a few > Windows users have reported performance regressions. Can you try > 4.3pre1? If the problem is not there, the curplit is almost certainly > the scrolling code. I have now tested jEdit 4.2final, 4.3pre1, 4.3pre2 and 4.3pre3. You are right, 4.2final and 4.3pre1 are OK, 4.3pre2 and 4.3pre3 show the delays! Manfred > Perhaps one of the new jEdit maintainers would like to take a look at > the scrolling code? I think it can all be ripped out, and the > JEditTextArea wrapped in a JScrollPane (which performs the bitblit > optimization too, however the code that does this is not public so I > could not reuse it in JEditTextArea directly without wrapping it in a > JScrollPane). > > Slava > > Manfred Usselmann wrote: > > Hi, > > > > I just tried to switch over to 4.3pre3 after using 4.2final for a > > long time. But on my system (Notebook with 650 MHz, Win XP, 512 MB > > RAM) 4.3 is nearly unusable for me, because editing is too slow. > > > > The problem is that simple editing tasks like moving the cursor, > > selecting text (with mouse or keyboard), etc. are executed slightly > > delayed. The biggest problem is with horizontal cursor movements. > > > > I disabled all plugins, problem remained. > > > > I'm back on 4.2final now. > > > > Regards, > > Manfred |