Lars Gullik Bjønnes wrote:
Abdelrazak Younes <[EMAIL PROTECTED]> writes:
| Abdelrazak Younes wrote:
| > Peter Kümmel wrote:
| >> Joost Verburg wrote:
| >>> Peter Kümmel wrote:
| >>>> Here a line from the scons log:
| >>>>
| >>>> cl /nologo /TP /EHsc /wd4819 /wd4996 -DHAVE_CONFIG_H
| >>>> -DQT_CLEAN_NAMESPACE -DQT_GENUINE_STR -DQT_NO_STL
| >>>> -DQT3_SUPPORT /IC:\Programme\zlib\include
| >>>>
| >>>> no optimization is enabled, but it links against the
| >>>> release Qt4 libs: QtCore4.lib QtGui4.lib.
| >>> I did enable all optimizations using CCFLAGS. With LyX 1.4 and Q../Free
| >>> I can scroll the whole Users Guide in 10 seconds, with LyX 1.5 and Qt4
| >>> it takes 30 seconds.
| >>
| >> Oh, now I understand, that's a difference between Qt3 and Qt4.
| >> From 10s to 30s that's really bad for on "upgrade".
| > I've said it multiple time but in case you didn't know, part of that
| > slowness is due to the fact that Key buffering is disabled in qt4.
|
| And I was wrong multiple times!
|
| I am applying this patch to my branch that will enable conditionally
| the keyboard buffering with the USE_KEY_BUFFERING macro.
The code you have in USE_KEY_BUFFERING is not key buffering, it is a
way to prune events that enter the queue too quickly. If f.ex. I keep
a key pressed, I expcet the insertion of the key to stop the moment I
release it. Same for scrolling, I expect the scroll motion to stop the
moment I release the PgDn key.
So there seem to have been some misunderstandings here.
Indeed. I thought it was a way to have a "fixed" number of PageDown
commands for a given period regardless of the drawing capability. I
thought that intermediate screen update would be skipped if its too late
for them. I got this impression by reading the mailing-list archive. It
was then explained that with this feature should speed-up the PageDown test.
If it is like you said then I don't think this feature is needed because
PageDown or ArrowDown always stop at the moment I release the key.
Thanks for the clarification,
Abdel.
