Peter Kümmel wrote:
Helge Hafting wrote:
Lars Gullik Bjønnes wrote:
| What is the problem that you are trying to solve here?
Is it my old pet? "Countinued scorrling after key-release"?
What was wrong with my patch from months back?
I tried this with todays svn.
The patch will indeed prevent scrolling after key-release,
so it fixes the pageup/pagedown case.
My patch eats the pageup/pagedown key events, but I don't
know what Lars' patch does.
It does not help for jumpscrolling using the scrollbar - LyX will
then overshoot a lot, if the scrolling work needs more than 100% cpu.
Attached a patch which is not mentioned to go in, but does it
at least fix the overshooting when scrolling by clicking in
the scrollbar?
Index: src/frontends/qt4/GuiWorkArea.cpp
===================================================================
--- src/frontends/qt4/GuiWorkArea.cpp (revision 18599)
+++ src/frontends/qt4/GuiWorkArea.cpp (working copy)
@@ -234,6 +234,7 @@
void GuiWorkArea::adjustViewWithScrollBar(int)
{
scrollBufferView(verticalScrollBar()->sliderPosition());
+ QApplication::syncX();
}
This fixed it perfectly!
No more overshooting. Performance was unchanged, not better, no worse.
I have the impression that "syncX" isn't too popular, but I wonder:
Will it really be a problem? The vast majority of fast machines
out there don't need 100% cpu to scroll. That means they
are idle between the scrollbar autorepeats. syncX is
probably not a problem for those?
A slow-display machine will scroll less (not slower - it was already
using 100% cpu, scrolling slower than the autorepeat anyway.)
So it will scroll less, because it don't overshoot.
I also tested "syncX" performance over the network. I used ssh
to get to a machine 20kM from here, and then ssh back to
this machine again where I ran the patched LyX. This sends the
graphics on a roundtrip 20kM back and forth - twice as bad as
just using a remote LyX. The scrolling was a little slower, but
it still stops perfectly. And the slowness wasn't too bad, not
something I mind when working remotely. ADSL has its limitations
anyway.
Please - if you can't come up with something better for 1.5.0, just
apply this.
It is simple, it works perfectly, and don't seem to kill performance.
Unless it has some other horrible drawback that didn't show up
in my tests. The overshooting certainly was horrible.
Helge Hafting
