GitLab

Patch by Ben Scholzen. Thanks!
Closes D779

The import/export is limited to what Krita already supported when
creating new brushes, i.e., not the whole GIH spec is implemented.

CCMAIL:kimageshop@kde.org
Implements T125

This reverts commit 3cea2871.

All the nice cleanups that Michael so painstakingly introduced are
undone, woe is me :-(

Instead of using a QImage, either full-sized or already down-sampled,
that then gets scaled again.

CCBUG:352781

This reverts commit dd81431d.

CCBUG:352781

Summary: Done with Google cppclean

Reviewers: rempt

Reviewed By: rempt

Differential Revision: https://phabricator.kde.org/D374

Instead of including it through ksavefile.h from kdelibs4support

Conflicts:
	krita/ui/CMakeLists.txt

Summary:
Allows importing/exporting local files in KisImportExportManager.

Importing/exporting .png, .jpg and so on were broken on my machine following the port to QUrl, because the to/from string functions became less forgiving.  For example QUrl("/home/miabrahams/test.jpg") is a no-go, it must be of the form "file:///home/miabrahams/test.jpg".  Similarly QFile(QUrl.toString()) will fail.

Reviewers: rempt

Reviewed By: rempt

Differential Revision: https://phabricator.kde.org/D356

Frame received is executed in the context of the image, so
we should decouple changing a state with a signal

Global defines:

  -DQT_USE_QSTRINGBUILDER
  -DQT_STRICT_ITERATORS
  -DQT_NO_SIGNALS_SLOTS_KEYWORDS

For libs and krita:

  -DQT_NO_URL_CAST_FROM_STRING

I cannot manage to fix all the places where a string is cast to
a QUrl in all of Calligra, though I fixed one egregious oopsie
in braindump.

I also would like the QString and QByteArray ascii cast defines,
but that's a huge battle!

The approach taken is a bit different from what's in the porting
scripts or in pigment: I've moved kis_debug to libglobal and adapted
the dbgKrita etc. defines to use qcDebug etc. I've kept the old
category numbers for now. I've also ported all places where we
used kDebug() etc. directly, and not a few places where qDebug() etc
were used directly.

This patch fixes multiple slowdowns caused my our background
working threads:

1) KisRegenerateFrameStrokeStrategy should *not* invalidate LOD planes
   (actually it doesn't write anything). So it should become a Lod0 stroke
   instead of a legacy one. Now it has a fake LodN buddy and suspend/resume
   strategies to mimic to Lod0 stroke.

2) KisAnimationCachePopulator should do the colorspace conversion in a
   background thread not related to the strokes. For bigger images (8k) or
   weird color spaces (Float32) the conversion might take up to 2-3 seconds
   and running the conversion in a context of a stroke will block the user
   from doing useful things all this time.

3) LodN planes should be synced from time to tine even when there are no
   Legacy strokes are present. The point is some brushes generate not
   identical results on different levels, so the two histories diverge
   with time. Now there is a special KisIdleWatcher class that runs a job
   when the image is idle (no strokes are running).

4) KisAnimationCachePopulator is also connected to its own populator
   stored in KisPart that helps it not to call barrierLock() every half
   a second.

Unless we can determine whether the image is still processing something,
we shouldn't suggest it is. This error message happens every time a
filter fails without setting an error message, which is misleading in
99% of the cases.

Next: check all our filters to properly set an error message.

And then we need to rewrite the filter system to be not so incredibly
complicated. Probably better reserve that for 3.x though.

CCBUG:347850