GitLab

0050-Fix-using-tablet-on-QML-widgets.patch is now broken, we need to
fix the bug again.

- Removed unused definitions and removed the unecessary information such as export path from
original authors system
- Saved the svg in plain svg format to reduce the size
- Svgs are now migrated to new inkscape dpi
- Changed the license to CC-BY-SA 4.0 like other svg in the repo earlier the license was proprietary.

Before this commit, curve would change semi-randomly in some cases
if you change from some specific sensors to some other specific sensors
(having a complex curve and clicking randomly should show a bug though)
when "Share curve across all settings" is selected.
This commit fixes that behaviour.

BUG:383909
CCBUG:410158

Before this commit, "Use the same curve" (precisely,
"Share curve across all settings") was done using
(1) rewriting values in sensors to the selected curve
(2) cached curves for all the sensors in case the user changes
their mind and wants the previous ones.

This behaviour brings confusion and is hard to debug, hence
the change.

This commit adds a new separate variable that keeps the
"common curve", so all sensors has their own curves.
There is no need for caching or restoring, it just needs
to be clear that when the Use The Same Curve checkbox is
checked, the value from the common curve needs to be taken
instead of the curve from the sensor in question.

Steps to reproduce:
1) Create a layer with colorize mask
2) Disable "Show key strokes" on the mask (important)
3) Select "Colorize Tool"
4) Click on the mask, to lazily activate key strokes
5) Close Krita -> Crash

The crash happened because of some weird update signals coming
after KisView was actually destroyed.

During the sprint we discussed that with painters and they said that
it would be much easier for them if all the layers would be visible
in the timeline by default. If the user can always hide a layer if
he likes.

CC:kimageshop@kde.org

The setting was present in KisImageConfig, but was never used in the
code itself.

There was a typo that loaded saturation from stored opacity.

CID:34704

CID:34707

CID:35124

CID:35317

Before this commit, if a preset had at least one random offset (X or Y)
in texture settings (and texture enabled), it was marked dirty as soon
as the user used this brush on the canvas. That was happenning because
random offset just means that the offset of the texture is set
to a new random one after every stroke. Setting a property to a new value
marked preset as dirty.
This commit ensures that in this case te preset is not set to dirty
(it just disables dirty notifications while the new offset is set).

BUG:406427

CID:35378

CID:35379

CID:35381

It leaks memory.

CID:35382

CID:35383

CID:35384

CID:248404

CID:248434

CID:248770

CID:248566

CID:248569

Even when we're loading a Krita 3 file with text, we won't
be loading ToC or bibliographies.

CID:248581

CID:248693,248678,248652

CID:248730

Variable anchor going out of scope leaks the storage it points to
if anchor->anchorType() == KoShapeAnchor::AnchorPage.

CID:248749

CID:248770

removeResourceAndBlacklist frees cs, so we shouldn't access it

CID:248787

CID:248835,248811,248799

CID:248853

CID:248855

CID:249012

CID:249013

CID:249014

(DIVIDE_BY_ZERO)12. divide_by_zero: In function call divide, division
by expression newDstAlpha which may be zero has undefined behavior.

CID:249018