GitLab

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"

Summary:
Currently a glitch exists where double clicking with a mouse while using the ctrl key-activated color picker causes the `KisAbstractInputAction::eventPos` and ``KisAbstractInputAction::eventPosF` functions to return an incorrect value, which then causes the ctrl-picker to return an undefined and unwanted color (usually black or white).

This patch fixes this bug by simply adding the `QEvent::MouseButtonDblClick` case to both switch statements, returning the correct position info for the abstract input events.

I believe this should close https://bugs.kde.org/show_bug.cgi?id=394396 . Although the user reported a crash associated with this bug, I was unable to reproduce any.

I also threw in some tiny formatting/style changes while I was in those files.

Test Plan:
Using a mouse, select the Brush Tool, hold the Ctrl Key to pop up a color picker, and double click.

Without this patch applied the picked color will often incorrectly be pure black or pure white, which seems to change depending on the zoom level.

After this patch, double clicking should not create any unwanted or unpredictable behavior from the ctrl-picker.

BUG:394396
Reviewers: #krita, rempt, dkazakov
Reviewed By: #krita, dkazakov
Subscribers: #krita
Tags: #krita
Differential Revision: https://phabricator.kde.org/D12953

BUG: 393396

BUG: 389045

PYTHONPATH should not be set to python3-binaries in
the environment when python2 is used often

Note: I have no idea whether the createtarball ruby script would
still add this code at the end of the top-level CMakeLists.txt
file...

Second note: to fetch the translations, run cmake with the following
option:

cmake ../krita -DKDE_L10N_AUTO_TRANSLATIONS=ON -D... This will
fetch the translations from subversion (so svn needs to be installed).
If you build master, it will fetch the right translations.

Not sure whether if you're on the stable branch, it will automatically
know that that's the stable branch...

You can also run "make fetch-translations" manually.

CCMAIL:kimageshop@kde.org
BUG:394381

OSX supports core profile only, so it doesn't report any extensions
for floating point texture formats. We should just use it directly if
openGL version is higher than 3.0, when floating point textures became
core profile.

Add tablet tester from Drawpile, with some changes.

Differential Revision: https://phabricator.kde.org/D12710

The source files are not yet modified.

This fixes some issues operating the touch docker and other QML parts
with touchscreen when using Qt LTS later than 5.9.3. See
https://bugreports.qt.io/browse/QTBUG-66718

Differential Revision: https://phabricator.kde.org/D12762

Summary:
This patch is part of task T8708, focused on further improving Krita's color picker.

>As of right now, Krita effectively has two near-duplicate paths when it comes to color picking code: the code for the dedicated Color Selector Tool (P) exists mainly in "kis_tool_colorpicker", while the code for the ctrl-key activated color picker that you can pull up while other tools are active [as well as the scratchpad's picker] exists mainly in "kis_tool_utils".

This patch addresses this issue by, as much as possible without some much bigger changes, unifying all of the core color picking functionality within the `KisToolUtils` namespace's `pickColor` function.

Essentially all of the duplicate code was removed from `KisToolColorPicker::pickColor`, which now calls into the same `KisToolUtils::pickColor` function that's used elsewhere creating a single place in the code where things like sampling radius and color picker blending are implemented. As a result, any change to the code in `KisToolUtils::pickColor` will now affect all of Krita's color pickers (the dedicated tool [P], the ctrl-activated picker, and the scratchpad picker) in a uniform and predictable way.

While this patch does solve the most of the code-duplication issue that existed in Krita's color pickers, there are still a few ways in which the dedicated tool and the ctrl-picker behave inconsistently. In my view, more extensive refactoring can be done in the future to create more uniform and predictable behavior across the various pickers, resulting in more color picking code being moved into `KisToolUtils::pickColor` from elsewhere. For now though, I wanted to focus on code-reuse improvements that don't have any user-facing design implications.

Also, some minor code-style cleanup was done here and there.

Test Plan:
Test the various color pickers (dedicated tool, ctrl-picker, scratch pad) for *mostly predictable and equivalent behavior.

*Certain aspects of the dedicated/ctrl picker can't be unified without some bigger changes, such as 'sample from merged/active only'.

Reviewers: #krita, #krita_abyss, rempt, dkazakov

Reviewed By: #krita, #krita_abyss, dkazakov

Subscribers: #krita_abyss, #krita

Tags: #krita_abyss

Differential Revision: https://phabricator.kde.org/D12941

Summary:
This patch adds timing functionality to the Animation Timeline Docker's "Insert N Keyframes" menu actions. Out of necessity, it also replaces the create-on-demand QInputDialog with a new TimelineInsertKeyframesDialog that was designed as a drop-in replacement that allows for getting more than a single variable of user input (i.e. frame count and timing) and which could easy be given more functionality in the future.

{F5845888}

The motivation behind this patch was to improve the "Insert N Keyframes Right/Left" action workflow by giving the animator control of frame timing. Previously, those actions worked by adding a number of immediately adjacent frames based on the user's desired number of frames to insert.

In animation, keyframes that are immediately adjacent to each other are described as being "on 1s", but other timings are also common - "on 2s", "on 3s", etc. - in which drawings are held for different amounts of time. Animations typically maintain a particular rhythm for a span of frames, but it's also very common for different parts of a single animation to switch timings.

This patch improves upon existing functionality to facilitate animating on 1s, 2s, 3s, and others. Now, a Krita user can specify a timing of 2 to quickly and easily insert keyframes "on 2s", a timing of 3 to insert "on 3s", and so on, allowing even longer timings if desired.

( Pair-programmed on a sunny Portland Saturday with @eoinoneill, of course! =] )

Test Plan:
1. Right-click a position on the animation Timeline docker.

2. Select "Keyframes > Insert N Keyframes Right" or "Keyframes > Insert N Keyframes Left".

3. A new dialog windows should pop up, asking for a number of frames and a timing.

4. Test different values as well as both the "Cancel" and "OK" buttons and check for predictable behavior in a variety of cases.

Reviewers: #krita, #krita_abyss, dkazakov, rempt, scottpetrovic, Bollebib

Subscribers: #krita_abyss, #krita, eoinoneill

Tags: #krita_abyss

Differential Revision: https://phabricator.kde.org/D12843

Windows support will come later as discussed in
https://phabricator.kde.org/D12838. To compile use the
-DENABLE_PYTHON_2 flag. Python 3 is still the default.

For the plugins to work correctly we need the tollius and pathlib
modules through pip.

Differential Revision: https://phabricator.kde.org/D12838

A KisPaintOp cannot create QWidget based objects because it doesn't
run in the gui thread.

BUG:392190

Special thanks to Peter Eszlari for finding the commit that
introduced the bug!

CCMAIL:kimageshop@kde.org

The original name was quite generic and didn't hint at
what it was meant to convey.

BUG:394015
patch by Jeroen Hoolmans, thanks!

BUG:394261
Patch by Jeroen Hoolmans, thanks!

BUG: 393586

(cherry picked from commit aa58837b)

The touch docker isn't that awesome on systems without a touch
screen and running the qml plugin saves cached compiled files
in the app bundle and that breaks signing. So disable this
until I've found out how to make Qt Quick on macOS save its
generated files in the user's home folder.

CCBUG:394188

Compare function should return false when comparing an item with itself

BUG:394068

CCMAIL:christoph-schaefer@gmx.de

CCMAIL:christoph-schaefer@gmx.de