Since df62a2c81b gb_DEBUG_CFLAGS
was passed to bridge_noopt_objects target for MSVC to track
down EH breakage on 64 bit architecture. It was erroneously
done unconditionally. Fix it to only pass when one of:
--enable-dbgutil
--enable-debug
--enable-symbols
is present.
Change-Id: I2728c30156541be418c88a0ea284c0c6a16abba5
Reviewed-on: https://gerrit.libreoffice.org/19744
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
...in code introduced with be1b03a4b0 "dock icon's
menu doesn't begin with separator anymore," causing CppunitTest_vcl_lifecycle to
fail
Change-Id: If6eb1f3a0f52d05284ac01f448888aaaa7523a98
Chose not to use lambdas instead because the number of arguments in these calls
would have made it way too verbose, hurting readability.
Change-Id: I8662a937ec10ca0fac6cd7e0d78b6268e8023ada
Reviewed-on: https://gerrit.libreoffice.org/19707
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: David Tardon <dtardon@redhat.com>
... to just assign a new value, which takes a significant amount of time
(~13% of ScTable::ValidQuery()) in ScRefCellValue::operator=() due to
its implementation using a temporary and swap.
Change-Id: Ia205850e10c5fa9083eec5fb7563d98561b52462
...at least recent Clang trunk ASan+UBSan build apparently adds instrumentation
code that made that (arguably optimistic) setup fail
Change-Id: I186a7abdfe5cc69e624583b1281d3989cefd7d53
(i.e., if any of --enable-dbgutil, --enable-debug, or --enable-symbols);
introduce gb_DEBUGINFO_FLAGS for the platform-specific flags
Change-Id: I4306fa30ced57d40d5b1206a0b26a886411118ef
(i.e., starting with an authority component); treating input starting with a
single slash (i.e., starting with an absolute path component) as a relative URL
instead of as an absolute pathname would cause e.g. CppunitTest_sal_osl_file to
fail
Change-Id: Ie340881974c5e9451ab7e0a9bfb21176b8f5666d
VCL Image/Bitmap/BitmapEx instances must not have static life-time
because then they will be destructed after DeInitVCL() and that
likely segfaults.
Change-Id: I3ff8d32de729c971b190028094cb4efe206395e2
There were two problems here:
1) ScTabView::getRowColumnHeaders() did not expose twip values directly,
but used ScRow/ColBar::GetEntrySize(), which does a twip -> pixel
conversion, and then converted it back to twip. Avoid this unnecessary
roundtrip.
2) ScViewData::ToPixel() trunaces the resulting float to an integer, so
if the result is e.g. 67.7 pixels, then Calc handled that as 67, but
gtktiledviewer rounded that up to 68, resulting in non-matching headers
for the rendered tiles.
Change-Id: Ie6ed1ea923a423d1526eeb235b7b87106fd2f20b
There was a subtle change in semantics, the earlier loop was only
iterating until aEnd, covering only unique elements of the vector.
This partially reverts commit c04fd82433
Change-Id: I75c5f1b343c3fa8855f6d7f8f706b9259bc9f6df
...expecified in File - Open too.
Change-Id: I616e66a296d7c44e5a8784b3a5f32f14ce7a668d
Note: Chrome and Firefox use back as well so IMHO it makes sense to do it this way.
Reviewed-on: https://gerrit.libreoffice.org/19567
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Norbert Thiebaud <nthiebaud@gmail.com>
As that's the unit we use everywhere else in the LOK API. Also, make the
ScGlobal::nScreenPPTX/Y calculation more precise, otherwise rounding
errors occur during the pixel -> twip conversion.
Example with the old precision: col height is 103 px, nScreenPPTY is
0.067, twips is 1537.3134328358208, convering it back is 102.487562189
px.
Example with the new precision: col height is 103 px, nScreenPPTY is
0.0667, twips is 1544.2278860569716, convering it back is 102.948525737
px.
Change-Id: I19f5285508ef0c751614d07969b3a7a037e7d1ec
