since we only care about appending to this container, and then
traversing it (normally once). So reduce the re-allocations that
std::vector requires
Change-Id: I206a7b82d9eefc1fa3762c4a03e7b5e21136951f
Reviewed-on: https://gerrit.libreoffice.org/30706
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
instead of returning a Primitive2DContainer from each method which we
are then going to immediately append to another container, pass down a
single container by reference which we can append to
Change-Id: I0f28a499d2ec54f7111a7044c30099767aa079e1
Reviewed-on: https://gerrit.libreoffice.org/30258
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
The problem is that the graphics cache only counts the size of the SVG
text, which is stored in SvgData::maSvgDataArray. However the
SvgData::maSequence may use a lot more memory, as it may contain
de-compressed bitmaps that are stored as base64-encoded PNGs in the SVG
text.
For example icon-themes/galaxy/brand/flat_logo.svg is 812 Ko but contains
60 Mo of bitmaps.
This may cause excessive memory usage and failure to export documents
due to OOM; according to valgrind massif, the bitmap buffers use 90% of
the heap.
Add a new interface com::sun:⭐:util::XAccounting, and implement
it in drawinglayer BasePrimitive2D. VCL SvgData can't access
drawinglayer via C++ directly so this looks like the best approach.
Change-Id: I5a7c3147733e23473c1decabed24c1f79d951c7d
Reviewed-on: https://gerrit.libreoffice.org/30669
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
The issue of 362d4f0cd4e50111edfae9d30c90602c37ed65a2 "Explicitly mark
overriding destructors as 'virtual'" appears to no longer be a problem with
MSVC 2013.
(The little change in the rewriting code of compilerplugins/clang/override.cxx
was necessary to prevent an endless loop when adding "override" to
OOO_DLLPUBLIC_CHARTTOOLS virtual ~CloseableLifeTimeManager();
in chart2/source/inc/LifeTime.hxx, getting stuck in the leading
OOO_DLLPUBLIC_CHARTTOOLS macro. Can't remember what that
isAtEndOfImmediateMacroExpansion thing was originally necessary for, anyway.)
Change-Id: I534c634504d7216b9bb632c2775c04eaf27e927e
See svg bug doc, which is processed quite slowly. Beyond needing faster
renderers, there is also demand to improve the handling of primitives
created by SVG import.
Conflicts:
drawinglayer/source/primitive2d/patternfillprimitive2d.cxx
vcl/win/gdi/gdiimpl.cxx
Change-Id: I10992a5746b8b2d6b50e3ee3fe415a035685c9ba
Generating primitives for chart visualisation can be moved to a
paralell executed task that loads the chart, thus speeding up
initial visualization. This is not possible for e.g. PDF or print
targets, only for edit visualization. On fallback, the replacement
images of the charts are used which are metafiles and have less
quality as primitives, but load quicker.
Change-Id: I68caa9e1bec50832bce535b5f54633d53cdef037
Isolated to a single Primitive2D class based on the AnimatedSwitch-
Primitive2D which does the specializing in one place. Buffers small
GIFs completely, handles 1st frame always buffered, huge GIFs get
animated by just playing he next frame.
To reach more with the current approach we would have to re-implement
AnimatedGIF import, replay it internally on a sys-specific Surface
and blit the current content (with alpha) to our display
Change-Id: I46c3325fa7936df73bea9a9284a0421f1475a34b
Reviewed-on: https://gerrit.libreoffice.org/26103
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
This try uses full 3D renderers working on the same ZBuffer
target, but are capable to render one stripe per thread.
This is rougher in granularity and uses multiple cores better
than the first try (see gerrit 24393) which was too fine-granular
being based on scanline render parallelization.
SecUred some more classes based on SdrPrimitive3D for multi-
theaded usage (places where local buffered stuff is done)
Change-Id: I4ddd5885ad41dd6432d0695e528818a86e427bfd
Reviewed-on: https://gerrit.libreoffice.org/24538
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
Reviewed-by: Armin Le Grand <Armin.Le.Grand@cib.de>
All DrawingLayer Atrributes used for ptimitives need a default
constructor. Since the o3tl::cow_wrapper is used in combination
with a static default incarnation of the impl class it is better
to define that. Also needed is a working isDefault() implementation
that compares the impl class to the static default. Added missing
stuff for this.
Change-Id: I98e9fee7343112cd979b972229423ac0ef3994a6
Reviewed-on: https://gerrit.libreoffice.org/24496
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Armin Le Grand <Armin.Le.Grand@cib.de>
The property stroke-miterlimit is transported to the renderers
via a new member mfMiterMinimumAngle in class LineAttribute
Several drawPolyLine methods are adapted. This patch does not
include changes in MetaAction. Presentation mode, printing, and
PDF-export is still wrong.
Corrected LineJoinMiter to LineJoinBevel in canvas, that s closer
to NONE. Removed DrawPolyLine method without MiterMinimumAngle
and adapted calls accordingly.
Change-Id: I6bcd24add5d85c4d9a39e3788e0682091c5fc9c4
Reviewed-on: https://gerrit.libreoffice.org/23946
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Armin Le Grand <Armin.Le.Grand@cib.de>
Reviewed-by: Regina Henschel <rb.henschel@t-online.de>
Regression from commit 2c91cb08d65cd35fa8ef6eaca3677aa82fb58cbe (better
drawing support for borders of different width, fdo#33634, 2012-04-04),
the problem is that previously the width of inner/outer double border
lines got rounded to integer values quite early, but after the commit
they are kept at a double precision for much longer, which needs pixel
correction in VclPixelProcessor2D.
Example: if the border with is 1.47, and the line gets moved by 0.2
pixels, then the inner and outer edge of the line will be 0.2 and 1.67,
which gets rounded to 0 -> 2 in the pixel processor. Previously the
input was rounded to 1, so moving by 0.2 resulted in 0.2 -> 1.2, which
got rounded to 0 -> 1. The result is that sometimes the line width is 1
pixel wider than expected.
Fix the problem by allowing VclPixelProcessor2D to request pixel
correction from BorderLinePrimitive2D. It wouldn't be possible to do
pixel correction only in VclPixelProcessor2D, as it has no idea what to
correct: it only gets polygons, so it has no idea if e.g. the top of a
polygon is the outer edge of a top border line or an inner edge of a
bottom border line.
Change-Id: I1971f3a952fbcdc598ab46c659e12d976c13cbe6
Reviewed-on: https://gerrit.libreoffice.org/24221
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
A ridiculously fast way of doing this is:
for i in $(pcregrep -l -M -r --include='.*[hc]xx$' \
--exclude-dir=workdir --exclude-dir=instdir '^
{3,}' .)
do
perl -0777 -i -pe 's/^
{3,}/
/gm' $i
done
Change-Id: Iebb93eccbee9e4fc5c4380474ba595858a27ac2c
Reviewed-on: https://gerrit.libreoffice.org/22224
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Chris Sherlock <chris.sherlock79@gmail.com>
<sberg> noelgrandin, just happen to look at b14224fe again; looks a bit scary to remove == or != from cases where both where declared
<noelgrandin> sberg, ok, I can revert that part
<sberg> noelgrandin, I guess that would be safer (there could be cases where now a different overload could kick in)
Change-Id: I5dc41c05dc4439d5adee0e5b3e0a9e1dfb9de3af
Reviewed-on: https://gerrit.libreoffice.org/22211
Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
Tested-by: Noel Grandin <noelgrandin@gmail.com>
using an idea from dtardon:
<dtardon> noelgrandin, hi. could you try to run the unusedmethods clang
plugin with "make build-nocheck"? that would catch functions that are
only used in tests. e.g., i just removed the whole o3tl::range class,
which has not been used in many years, but htere was a test for it...
<noelgrandin> dtardon, interesting idea! Sure, I can do that.
Change-Id: I5653953a426a2186a1e43017212d87ffce520387
Reviewed-on: https://gerrit.libreoffice.org/22041
Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
Tested-by: Noel Grandin <noelgrandin@gmail.com>
rather than uno::Sequence, since we modify this container a lot
and uno::Sequence is expensive to update
Change-Id: Id5bc5171cbc4b90b243e6dda6d572f21b3bdf00d
