Am 02.06.2010 um 12:08 schrieb Jean-Marc LASGOUTTES: > José Matos <jama...@fc.up.pt> writes: > >> Hi, >> I had this problem before but being busy I did not report it. >> >> One of the features of F-13 is >> http://fedoraproject.org/w/index.php?title=UnderstandingDSOLinkChange >> >> This means that lyx does not compile because when linking with qt we need to >> pass -lX11 since qt uses symbols from X11. > > This bug is also known in debian > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=555580 > > It is not the reason. With the new DSO thing linking against qt does > _not_ require to link against X11, and this is actually our problem. We > explicitely use X11 calls for some weird event ignoring in > GuiWorkArea.cpp, which fixes AFAICS the following bug > http://www.lyx.org/trac/ticket/3320 > > Before trying to add the -lX11 in autotools, let's see what we can do > about the actual code. It reads: > > // do nothing if there are other events > // (the auto repeated events come too fast) > // \todo FIXME: remove hard coded Qt keys, process the key binding > #ifdef Q_WS_X11 > if (XEventsQueued(QX11Info::display(), 0) > 1 && ev->isAutoRepeat() > && (Qt::Key_PageDown || Qt::Key_PageUp)) { > LYXERR(Debug::KEY, "system is busy: scroll key event ignored"); > ev->ignore(); > return; > } > #endif > > > Firstly, there is a very amusing detail: as I read it, the test for > Qt::Key_PageXXX does nothing :) Since nobody complained about it, and > since explicitly testing for hardcoded keys is tasteless, I propose to > remove this part. > > Secondly, XEventQueued can be replaced with > QApplication:hasPendingEvents, also there is a slight difference, the > later calls XPending: > > The XPending function returns the number of events that have been > received from the X server but have not been removed from the event > queue. XPending is identical to XEventsQueued with the mode > QueuedAfterFlush specified. > > So what do these mysterious flags mean? > > If mode is QueuedAlready, XEventsQueued returns the number of events > already in the event queue (and never performs a system call). If > mode is QueuedAfterFlush, XEventsQueued returns the number of events > already in the queue if the number is nonzero. If there are no events > in the queue, XEventsQueued flushes the output buffer, attempts to > read more events out of the application's connection, and returns the > number read. If mode is QueuedAfterReading, XEventsQueued returns the > number of events already in the queue if the number is nonzero. If > there are no events in the queue, XEventsQueued attempts to read more > events out of the application's connection without flushing the output > buffer and returns the number read. > > For information, we use mode 0, which is QueuedAlready. It seems to me > that using QueuedAfterFlush is a better idea, so I think using > hasPendingEvents is good enough. > > This leads to the following patch. Now is looks like the code should be > useful on other platforms too. Is this problem with page down also > present for Mac OS or Windows? I would propose to use the code on > all platforms.
I've tested 1. the original code, 2. the patched one, 3. with removed ifdef Q_WS_X11. At the first glance there is equal behavior for all 3 versions. When scrolling fast with page-down key pressed I see "isAutoRepeat: 1" in debug output. When I release the key the scrolling stops immediately. For MacOS your patch makes no difference, because Q_WS_MACX is defined instead of Q_WS_X11, right? But I would say it should work with Q_WS_MACX too. Stephan
