The problem was that SCA_Joystick::pAxisTest() was using shorts, and tried to store abs(MIN_SHRT) in a short. However, on most systems MIN_SHRT == -32768 and MAX_SHRT == 32767. This means that abs(MIN_SHRT) > MAX_SHRT, and thus the short would overflow.
- Old issue: on scrolling button views, tooltips could open or stayed open.
- New fix: alt+swipe now changes button values again
- Removed test print, from WIP code project.
sculpt_multires_active() now returns NULL if dynamic topology is
enabled. Fixes bug #33718:
projects.blender.org/tracker/?func=detail&aid=33718&group_id=9&atid=498
since we fill it later anyway. Usually OpenGL does color + depth buffer
concurrently so this probably won't have any noticable effect. Still
better be pedantic about it in case we do earn some performance out of
it. Added alpha component in sky color to make sure it is set to zero in
the framebuffer too.
Trackpad zoom (swipe + CTRL) direction was inverted compared to MMB-drag
or scrollwheel usage. In the 3D viewport it was OK, in all others not.
Now the same physical gesture maps identical to zooming everywhere. Or to
recap (with blender factory settings)
Zooming in:
- MMB-drag, move mouse towards screen
- Scroll wheel, move finger towards screen
- Magic Mouse, move finger towards screen
- Trackpad 2-finger swipe: move fingers toward screen.
To make this extra confusing: this is only consistent if you set your system
to inperpret trackpad swipes as "inverted" (pan view left = swipe to right).
This is a typical default, although Apple wants you to call this "Unnatural" :)
Next commit will be testing on laptop if all pinch gestures zoom consistent.
And following to that, a sensible user preference to map trackpad use for
Blender yourself, to invert system defaults again. :)
Blame and thanks goes to Sebastian Koenig, for his perseverance on getting this
solved :)
This will only make object display with proper shape in textured
view, but all materials and textures will be replaced with default
gray color. There's currently no better way to deal with textured
display when dynamic topology enabled because of all UV/tfaces are
clearing when enabling dynamic topology sculpt.
Anyway, better to display gray object with proper lighting in this
case rather than not update object's shape during sculpt.
Proper solution will be possible once CD layer will be preserved
by BMesh log.
without hurting quick texture painting
- ED_view3d_draw_offscreen will now output buffer with
transparent alpha, if sky needed it should be alpha-undered
later.
- ED_view3d_draw_offscreen_imbuf now accepts alpha mode as an
argument which could be either R_ADDSKY or R_PREMULALPHA
- OpenGL render and sequencer's opengl preview will now reflect
scene's Alpha Mode
- Quick Edit will use OpenGL with transparent alpha mode
The current C callback functions are too simple allow a straightforward implementation, in particular they don't receive the PropertyRNA pointer itself as an argument, which means the callback cannot directly access the PropertyRNA's py_data pointers which store the python function objects. For this reason a second runtime variant of these callbacks has been added. It is only used for runtime callbacks and not in makesrna, but otherwise works the same way.
This sub-type is actually *only* needed for the "text" property of UI rna api (maybe we should rename it to "PROP_PY_TRANSLATE", as it is anyway only 'active' during conversion from py string to RNA string property...). In fact, I think it should only be used in RNA func properties anyway, as it stores the translated string into the property, it should only be used with "one time" RNA stuff...
purpose of some of these features
* ClampTo Constraint, Target Object field: now mentions that it is for curve
objects only
* Dopesheet Editor Modes: now describe what data is editable in each mode. The
wording may be a bit clumsy here, but (apart from Mask Editor one, which I'm not
entirely sure about) should at least be technically correct
in a timeline view would show up search results in other editors (e.g. 3D View).
This only really works where there is a timeline view as:
1) The modal interactive stuff is built on the assumption that a suitable 2D
view mapping exists
2) The operator is not "registered" (i.e. no props display after usage, and
doesn't get shown in undo) as it's more "view state" than "scene data"
Rewrote poll callback to handlethis properly
As of r53332, give_current_material() won't accept a 0 index and return NULL. CreateMaterialFromBlenderObject() was using index 0 to just grab the first material. Since this now returned NULL, it would use default settings. We'll just ask for index 1 now instead.
In fact surfaces should be displaying the same way in textures and solid shading,
but they used to change a way they're displaying by draw_textured_end not resetting
light model properly, leaving two-sided shading for all further objects draw.
Additional changes:
- Made mipmapping operate with unsigned short instead of char
which allowed to eliminate extra division by 255, so prevision
should be a bit better now.
- Actually, this is not real unsigned short range, but it's a
range of 255*255 which is more convenient for mipmapping, so
made conversion functions private for scaling.c
Not sure it worth making this functions operate in 65535
range, for now current behavior seems to be just fine.
Regression was caused by alpha premul cleanup commit and the reason
of slowdown was uchar <-> float conversion which is slow.
Replaced with uchar <-> int conversion which seeps to be accurate
enough and mostly eliminates slowdown.
Slowdown was easy to notice when movie clip is used for 3d vierport
background and undistortion is enabled. In this case every frame
will re-calculate mipmaps.
It's still a nit slower than mipmap generation before cleanup
commit, but couldn't think about extra boost here atm.
- Trackpad swipes now behave same as scrollwheel for listview scrolls
(disabling 2d view scroll when mouse over)
- Added back 2.4 debug print for glGetError()
Only useful for developers - to check what goes on when ogl messes up.
- Made more clear print for read factory default. It's not error :)
