Commit by coppens :: r903945 krita/trunk/koffice/krita/image/kis_paint_layer.cc:

When the layer has been moved after painting on it with a temporary target, the projection will not have moved. So move the projection as well, so
that we replace the right part of the projection with the paint device's data.

Commit by coppens :: r903924 libs/trunk/koffice/libs/pigment/tests/ (3 files):

Port the RGB8 code to the QTest framework and enable it. Add a new test to see if the copy composite operation works. (Disabled the old, unported
tests for now)

Commit by coppens :: r903611 krita/trunk/koffice/krita/plugins/paintops/libpaintop/ (3 files):

Somewhat more proper 'fix' for the bug referenced in r857756 (and revealed by the fix for bug 178599). The problem with that brush seems to have been
that it used a brush with brush selection modes for tiltx and tilty, which we do not implement/support yet. However, this brush had 0 subbrushes for
these tilt dimensions, and the previously used fallback mode for tiltx and so was 'Incremental', which does a modulo operation with the number of
subbrushes (in this case 0 -> division by 0).

CCBUG:178599

Commit by coppens :: r903021 krita/trunk/koffice/krita/plugins/paintops/libpaintop/ (4 files):

Make the clone functions for brushes much simpler. Move the actual cloning logic to the copy constructor, and make them actually work. This fixes the
cloning of image and pipe brushes. Fixing the cloning unfortunately shows a bug in the parasite loading code introduced a while back (r857756), so
that you won't actually see the brush changing. I'll fix that after I find out which brush needed that commit to be fixed.

BUG:178599

Commit by coppens :: r884585 krita/trunk/koffice/krita/image/tiles_new/kis_tileddatamanager.cc:

Detach the tile before overwriting it in setting a new extent. Fixes a bug in cropping (as verified by the unit tests). Also don't doubly lock the
tile here, since after detaching the tile data might be different...

Commit by coppens :: r884581 krita/trunk/koffice/krita/sdk/tests/testutil.h:

Actually set the right coordinates for the failing point (makes debugging output more sensible)

Commit by coppens :: r879187 krita/trunk/koffice/krita/ui/kis_histogram_view.cc:

Don't try to calculate log(0) when drawing logarithmic histogram plots. Fixes the weird problems where almost everything was drawn in
black (and where in my test image the 0-value of green was obviously wrong as well). I can only wonder how this ever worked in 1.6 :-) (perhaps Qt
just didn't draw lines with those ridiculous coordinates?)

Commit by coppens :: r879179 krita/trunk/koffice/krita/plugins/viewplugins/histogram_docker/ (histogramdocker.cc histogramdocker.h):

Show the right channels in the histogram docker when Krita starts

Commit by coppens :: r879163 krita/trunk/koffice/krita/plugins/viewplugins/histogram_docker/histogramdocker.cc:

The histogram docker should really only show color channels. Makes more sense, I think

Commit by coppens :: r879130 krita/trunk/koffice/krita/plugins/viewplugins/histogram/wdghistogram.ui:

Histogram view dialog can now differentiate between linear and logarithmic display of the histogram (the Qt widget lost a signal during Qt3-Qt4
conversion, I guess...)

Commit by coppens :: r879089 krita/trunk/koffice/krita/plugins/viewplugins/histogram_docker/ (3 files):

Histogram docker code paths are used again, but there's something amiss with the visualisation part (try switching the displayed histogram to Lab,
you'll see a mini histogram being shown...)

Commit by coppens :: r878794 krita/trunk/koffice/krita/plugins/paintops/defaultpaintops/smudge/ (2 files):

Let the smudge paintop work somewhat again. You can't actually *see* it work, since there seems to be something wrong with the rate widget magic.
Manually setting the opacity in the code there to, for example, 127, seems to make it work nicely. I'll leave that to someone more skilled with the
widget stuff to fix.
It somewhat feels a bit 'chunky' when you smudge, which is odd. But at least it does something now.

Commit by coppens :: r877240 /trunk/www/areas/koffice/krita/gallery.php:

Spelling fix as requested by m4v

Commit by coppens :: r806040 krita/trunk/koffice/krita/ (18 files in 4 dirs):

Long-awaited next commit for the next-generation tile system of Krita. Took a long time to code it all in my spare time, despite the relative
shortness of the patch. It's rather evolutionary from the last mass-commit, in that the main change is a nice refactor of the shared tile data (I
like it a lot better than it was before, even though it could still be better). Has 2 (or 3) known issues at the moment:

• At program/testcase exit, the tile swapper will trigger an assertion. This is because I'm more assertive than before ;) Basically, Krita seems to be leaking some memory, and the tile swapper wants all tiles to not leak at all (an example of a fix of such a leak was the recent work on the update strategies). You can filter those pseudo-failing tests out by uncommenting the relevant asserts and deletes in KisTileSwapper::~KisTileSwapper
• Some KisFilterTests sometimes trigger another assert during program exectution. Will be investigated...
• While using my code to play with Krita, I sometimes seemed to get a random assert during painting or so, cause unknown (perhaps it even got fixed together with some other bugfix?) I'm sure Boudewijn, Cyrille and even Pippin will love that it I'm committing this :-) I guess they can celebrate it together with their fellow LGM'ers there in Wroclaw :-)

Commit by coppens :: r777171 krita/trunk/koffice/krita/ui/kis_qpainter_canvas.cpp:

Seems to fix the crash on startup I get

Commit by coppens :: r768356 krita/trunk/koffice/krita/image/kis_random_accessor.cpp:

Whoops, seems like I forgot to commit a file.

Commit by coppens :: r723818 /trunk/www/areas/events/info/conference2008/ (contact.php index.php menu.inc site.inc):

<CRAEW> BCoppens: could you just make the changes please or I will do it myself
<CRAEW> BCoppens: as head organiser of aKademy I sometimes have to make critical decisions, and Bart (Steccino) and I decided to use aKademy
<CRAEW> just everywhere where there is written Akademy change it in aKademy
Marketing people, get your act together and your people in line please, I really have better things to do than this...
CCMAIL:kde-ev-marketing at kde dot org

Commit by coppens :: r720277 /trunk/www/areas/events/info/conference2008/ (contact.php menu.inc):

• At least be somewhat internally consistent with the currently chosen spelling of Akademy 2008
• The old conferences were really spelled aKademy

Commit by coppens :: r720276 /trunk/www/areas/events/info/conference2008/ (contact.php index.php menu.inc):

Add contact information to the Akademy 2008 page. I am not going to add the other suggested pages just yet (Sponsoring, Press, Accomodation, Conference Programme), since I don't have any
senseful contents for it yet. Better wait until it is ready, than to have an unfinished page displaying 'You can download the brochure (coming soon)'.

Note furthermore that, thanks to the cool and easy archiveability, you can easily look at the early versions of the aKademy 2007 site for inspiration, through
http://websvn.kde.org/trunk/www/areas/events/info/conference2007/ and then clicking on the filenames to browse through their history.

CCMAIL: bart dot cerneels at gmail dot com

Commit by coppens :: r703563 krita/trunk/koffice/krita/image/tiles/kis_tileddatamanager.cc:

Fix the unit tests for planar separation code of the paint device. The issue is very simple: the code was copied from read/writeBytes. Those functions apparently are written for the specific case that the result of read is written on a paint device at the same coordinates as the coordinates of where it was being read (modulo TILESIZEs). Of course, that fucks up if you read and write at different offsets.
Fix it by forcing to read a single row at a time in the planarRead/Write code (yes Boudewijn, that makes it slower, but I can't help that ;-))