GetBitCount() works differently for VirtualDevices. GetAlphaBitCount()
is really only used by VirtualDevice, so moved functionality from
OutputDevice to VirtualDevice.
Change-Id: Ic00e32f1fa385542bcce8c9475f0ea5eb9a077f9
Reviewed-on: https://gerrit.libreoffice.org/8722
Reviewed-by: Chris Sherlock <chris.sherlock79@gmail.com>
Tested-by: Chris Sherlock <chris.sherlock79@gmail.com>
OutputDevice::EnableRTL() is a bit of a mess. It uses a runtime
variable to see if it is using a VirtualDevice, and it uses a
dynamic_cast to see if the object is a Window or a Control!
I have made it virtual and moved the knowledge of class specific
functionality from OutputDevice to VirtualDevice, Window and Control
as needed. OutputDevice::EnableRTL() functionality is then called.
Also: small formatting change to outdev.hxx, also included a note
that WindowImpl is a pimpl in window.hxx.
Change-Id: I44b66601c4457fb2e0bbc1014fb7acf8f6942f80
Reviewed-on: https://gerrit.libreoffice.org/8721
Reviewed-by: Chris Sherlock <chris.sherlock79@gmail.com>
Tested-by: Chris Sherlock <chris.sherlock79@gmail.com>
The vertical space of the options dialog is to small for the new
security page in German translation. The last entry is not visible
and the button could not be reached.
Solution: I added more vertical space to ROW_3 which is the postion
of the separation line. I added the hight of a button and of the line.
This fixes the issue with missing space for the last entry in section
especially for the German translation.
Change-Id: Icacfd4587d605e56438f3d7207ce5d257f313072
Reviewed-on: https://gerrit.libreoffice.org/8735
Reviewed-by: Björn Michaelsen <bjoern.michaelsen@canonical.com>
Tested-by: Björn Michaelsen <bjoern.michaelsen@canonical.com>
I occassionally get lockups in IceProcessMessages() called from QtCore,
I'm actually not exactly sure why, as theoretically two connections
from one app shouldn't be a problem, but since LO does its own
session handling, there's no need to the KDE/Qt code to be involved,
so prevent it from connecting to the session manager altogether.
Change-Id: Iebe20d4cb5403e5fea8bd5d8c1f69b62d1c2907b
With the document from fdo#75622, this saves 3775 calls and leaves only 13.
e586fe4585 removed avoiding the call
to LayoutTable(), which made loading slow. I checked that the doc from that
bugreport still works, so if very original code was correct in avoiding
the call sometimes, this should be ok too.
Change-Id: Ia80f974d4497e5cb04612331527eb87b579ddb76
The problem was that in
writerfilter::ooxml::OOXMLStreamImpl::lcl_getTarget(), we went over the
list of all hyperlinks for each request. Instead, let's do it once, and
in the remaining cases just look up the result from a map.
Numbers before on my machine (ms, load time):
2215, 2243, 2205
After:
1362, 1358, 1358
So that causes about 39% speedup for a non-debug build.
Change-Id: Ib4718abbe834c5ba49a85469148b656e3c808041
...the calls to getAccessibleChild(0) and getHeaderBar(BBYTE_COLUMNHEADERBAR)
ultimately all ended up at
AccessibleBrowseBox::implGetHeaderBar(BBTYPE_COLUMNHEADERBAR).
Change-Id: Ie25659e19d0d1ce2dcde2d1440ecc69b4fa265b0
The isEmpty could be redundant, since the first test for length in
the old code seemed to be just a pre-condition for evaluating the
expressions in the second one.
Change-Id: I0d1838d5eb143196a0d7be0cbadf9f0a59cff71b
