GitLab

BACKPORT:krita/4.2

When KisGuidesManager's setUnitType() is called when the mouse is
dragged onto the canvas, slotUploadConfigToDocument() will upload
the guides config that includes the newly created guide, resulting
one additional unexpected "Add Guides" command. The current
implementation will guarantee one undo command is created only
after the release of the mouse/finger/pen.

KisChangeGuidesCommand is now post-exec, which means the first redo
will be ignored. The command will not be created during dragging.
Instead, only when the mouse/pen/finger is released will the command
be created. The commands will not be merged.

Any set* function of KisGuidesManager will upload
the config to KisDocument.

Undoing the removal of the last assistant will not make the
removed assistant show up again. This is due to the removal
of the last assistant in `KisPaintingAssistantsDecoration`
will `setVisible(false)`, but the undo command has yet to
revert this.

This commit involves API changes in
`KisPaintingAssistantsDecoration`.

These properties will be shared between all clones of some assistant
using QSharedPointer.

Adding/removing/editing assistants will now trigger clones of the
assistant list. The undo commands will replace the current
assistant list with the saved ones for undo/redo.

Only adding/removing/moving a guide will trigger the creation
of KisChangeGuidesCommand. Only consecutive movements, or creation
then movement, or movement then deletion of the same guide
will trigger the merging of the commands.

CCBUG: 361012

BUG:408146

BACKPORT:krita/4.2

BUG:408113

BUG:359453

SvgContext spilled caused by copying pointer instead
of copying values for stroke.

All TestSvgParser pass. Attached bug files render as expected.
should be safe to port to krita 4.2

Bug:407683

BACKPORT:krita/4.2

BUG:404321
BACKPORT:krita/4.2

BUG:408002
BACKPORT:krita/4.2

In my HDR branch I have fixed Qt's bug of not-emitting
dockLocationChanged() for undocked state (the patch is accepted into
Qt 5.13 https://codereview.qt-project.org/c/qt/qtbase/+/252182).

Our code exploitated this bug, which should be fixed now :)

BUG:407916
BACKPORT:krita/4.2

Using the stylus' unique id to set the preset allowed users to
have different presets for e.g. an art pen and a normal stylus.

However, we are seeing random stylus id's for the same actual
stylus being reported, which means that this feature is completely
broken.

BUG:407659
BACKPORT:krita/4.2

GCC now says:

cc1plus: warning: -Wabi won't warn about anything [-Wabi]
cc1plus: note: -Wabi warns about differences from the most up-to-date ABI, which is also used by default
cc1plus: note: use e.g. -Wabi=11 to warn about changes from GCC 7

In case of conflict in i18n, keep the version of the branch "ours"
To resolve a particular conflict, "git checkout --ours path/to/file.desktop"

==23688== 115,060 (224 direct, 114,836 indirect) bytes in 2 blocks are definitely lost in loss record 12,891 of 12,945
==23688==    at 0x4C2E68F: operator new(unsigned long) (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so)
==23688==    by 0x129083D3: QuaZip::QuaZip(QIODevice*) (in /usr/lib64/libquazip5.so.1.0.0)
==23688==    by 0xA7DC821: KoQuaZipStore::KoQuaZipStore(QIODevice*, KoStore::Mode, QByteArray const&, bool) (KoQuaZipStore.cpp:69)
==23688==    by 0xA7D5DDF: KoStore::createStore(QIODevice*, KoStore::Mode, QByteArray const&, KoStore::Backend, bool) (KoStore.cpp:110)
==23688==    by 0x90ED6DE: KoGamutMask::loadFromDevice(QIODevice*) (KoGamutMask.cpp:245)
==23688==    by 0x90ECEE5: KoGamutMask::load() (KoGamutMask.cpp:211)
==23688==    by 0x89B4BB3: KoResourceServer<KoGamutMask, PointerStoragePolicy<KoGamutMask> >::loadResources(QStringList) (KoResourceServer.h:203)
==23688==    by 0x89AC270: KoResourceServerProvider::KoResourceServerProvider() (KoResourceServerProvider.cpp:142)
==23688==    by 0x89AC6BB: Holder (KoResourceServerProvider.cpp:156)
==23688==    by 0x89AC6BB: innerFunction (KoResourceServerProvider.cpp:156)
==23688==    by 0x89AC6BB: operator QGlobalStatic<KoResourceServerProvider, (anonymous namespace)::Q_QGS_s_instance::innerFunction, (anonymous namespace)::Q_QGS_s_instance::guard>::Type* (qglobalstatic.h:134)
==23688==    by 0x89AC6BB: KoResourceServerProvider::instance() (KoResourceServerProvider.cpp:160)
==23688==    by 0x5DCE9EB: KisApplication::loadResources() (KisApplication.cpp:287)
==23688==    by 0x5DD2C4F: KisApplication::start(KisApplicationArguments const&) (KisApplication.cpp:425)
==23688==    by 0x40698B: main (main.cc:524)

BACKPORT:krita/4.2

BACKPORT:krita/4.2

Before this commit, some .kra files weren't displayed properly after
loading - some were showing just transparent canvas, some displayed
properly some parts of the canvas while others were missing.
This commit should fix those issues.

For transform masks we shouldn't use exactBounds(), which includes both
transformed and original rects, but a specially crafted sourceDataBounds()
method that includes only source data.

BUG:407359
BACKPORT:krita/4.2

BUG:374748
BACKPORT:krita/4.2

It has to be this way to avoid being covered by tool windows, 
and before changing to nonModal as Tool window, a workaround
 for shortcuts must be made. 

As a Tool window the menu is not shown, making some edition
options unavailable.

BUG:407878
BACKPORT:krita/4.2

BUG:407804
BACKPORT:krita/4.2