Blender Git Loki

Patch #3365, Toolbox from Tuhopuu
Patch prvovided by Guillermo, code was - afaik - from Rob Haarsma.

This changes the toolbox (space menu) to have the first level aligned
vertically. Works much easier that way, and since the items open either
left or right, it doesn't flip order of the contents for it either.

To allow people to test (and to compare) it's a user menu setting (in
View & Controls, "Plain menus"). I've turned this on by default though,
since I propose to not have it a user setting. User setting can be
removed later.

Fixed two bugs in patch:
- if saved in user settings, first time usage of this toolbox opened in
wrong location
- Button for "plain menus" was writing a short in an int
(causing this new menu not to work for big endian systems)

As a bonus I've added the long wanted hotkey support for opening and
closing sublevels of pulldowns with arrow keys!

I didn't add the commenting out of correcting pulldown menu order, which
is based on location of the originating button in the UI. This uncommenting
didn't solve anything, since button definitions itself can be flipped too.
(Example: the data brose menus in top bar need to be corrected).

I can imagine the order flipping is sometimes annoying, but it still has
reasons to be there;
- the most important / most used items are always closest to the mouse.
(like opening properties panel, or "Add new" for material.
- it follows muscle memory and 'locus of attention' (mouse position).
- menus are configured to open to the top for bottom headers, and to the
bottom for top headers. We can expect the UI is configured consistantly
for headers, so in general the menus will appear consistant as well.

Where menu flipping fails is especially for alphabetic listings, like in
the menu button of fileselect. However, that one should be configured to
open by default to the bottom, so ordering is consistant as well.

If people like to check this themselves; uncomment the lines in the top
of the function uiBlockFlipOrder() in src/interface.c

-- fix bug which used unitialized pointer in M_Effect_New.

Removed static from shade_buttons_change_3d declaration and included
it in BIF_interface.h for general use in other files

Add a call to buttons_change_shaded_view() or something in 'assign material
index'

Removal of warning messages, unused variables, and casting errors.

Finally switched to porting UI stuff from tuhopuu!

This commit is based on the patch & cool design work of Matt. It includes
the new Lamp drawing style, and replaces the Object center dots with a
similar styled OpenGL drawn dot.

Important side-note is that removing the old glDrawPixels() for centers or
lamps will not only make Blender faster, but also prevents crashing on a
couple of cheaper 3d cards (as reported for S3 and Intel on-board cards)

Notes:
- The new default only draws Object centers when selected or active. If
you like to see them always, use the View Properties Panel. You can also
save that in the .B.blend
- The size for centers (and lamps) is in the User settings "View & Controls"
- Unselected Lamps, and their offset lines from zero Z, are drawn in a new
Theme color

Changes and additions in Matt's patch:
- Lamps and centers are drawn fixed size, in pixels. Also the 'sun' lamp
draws screen aligned now.
- Center dots now also draw in blue to denote Library linkage or to show
that an Object has been linked to other scenes.
- When objects are empty (no vertices) they will always draw a center dot.
Otherwise these objects would never be selectable anymore!
- Added theme setting for center size, and initialization
- Removed the old redundant code for drawing centers
- Cleanup of drawing routines, made center dots faster
- Started removing calls to glBlendFunc(). Regular alpha drawing should
become standard, and the (very) occasional exception should return this
to default after usage.

-- added Object.protectFlags attribute; let user get/set transform lock bits

-- fixed bug with returning NULL from Object_CreatePyObject() on MacOS/X.
The code prior to the last commit worked... but can't see the differene

Bugfix: in Sequencer, an active image/movie strip also allowed to add Ipos,
causing Blender to crash. (reported by Levon, thanks!)

Bugfix: InfoWindow, pulldown menu said "Dump 3D window", whilst this can be
any window type... renamed it to "Dump Subwindow" next to "Dump Screen".

- bug fix of bug #2923
bug was reported by ideasman, thanks! This bug fix also solve problem, when
MetaBall is parented to some zero size object.

Bug #3395

Target-less or Auto-IK didn't insert keys for all bones when using
the Auto-key option.

Two errors in old style (not relative) Shape Keys;

- they didn't even execute!
- you could not grab them in IpoWindow

Fix: When inserting keys for Forcefields, used for particles systems, you
could not move the object anymore (since the ipo got executed).
Solved with copying original Object data in the Deflector Cache.

New particle option "Unborn" and "Died" didn't work for shaded halos yet.
Thanks Chris Burt for the poke!

- bug fix #3380, patch provided by Ken Hughes, thanks

Fix in Colorband: the "B-Spline" interpolation didn't extend well, when a
marker wasn't on the first or last possible position. Caused by clipping.
As bonus; added Cardinal interpolation option too, which is just that
little bit different! (Cardinal goes through the controlpoints, bspline not)

A year-zero bug, and actually another issue with auto-handles for Ipo
curves. On sharp 'peaking' curves the handle was calculated
using both X and Y distance. This could result in overshooting.

New code only evaluates the X distance, resulting in much more moderate
sized handles.

Thanks Gabio for the demo file!

-- add support for new particle effects

-- fix small problem with Ipocurve.recalc(); recomputed handles but did not
re-sort control points (also added reminder in BezTriple doc that script
users need to call recalc themselves)

-- Fix for the "LIB ERROR: base removed" problem; BPy Object was decrementing
us.id when objects were destroyed but not always incrementing when
created. The intent of modifying us.id is to make Python a "user" of the
data so it persists even when it is deleted from Blenders UI. The original
commit was unintentional but Ton thought the idea was OK.