- define __BIG_ENDIAN__ or __LITTLE_ENDIAN__ with cmake & scons.
- ENDIAN_ORDER is now a define rather than a global short.
- replace checks like this with single ifdef: #if defined(__sgi) || defined (__sparc) || defined (__sparc__) || defined (__PPC__) || defined (__ppc__) || defined (__hppa__) || defined (__BIG_ENDIAN__)
- remove BKE_endian.h which isn't used
- remove saniy checks from blf_font.c, the callers now check instead.
- move duplicate code into defines (may move into static functions).
- move kerning checks into const values set at the start of the function, rather then checking on every character.
patch from Damir Prebeg with some edits.
Also made it so resizing the console view keeps the lower part of the text in view (could be annoying when you needed to scroll because of a resized view).
glGetIntegerv(GL_TEXTURE_2D_BINDING_EXT, &cur_tex);
... with shadow enabled glGetFloatv(GL_CURRENT_COLOR, color) was called twice per character as well.
Now only call glGetFloatv(GL_CURRENT_COLOR, ...) once per string and only when drawing with shadow or blur, texture bind is stored in the font.
Gives 8% overall FPS speedup when displaying heavy UI in my test.
- use vertex array for drawcircball()
- add circball_array_fill() and call from drawcircball().
- for object center's rather than drawing 2 circles, create the array and reuse it.
geez, who coded the font object? or even worse, who did the dome code?
Don't coders talk? tsc tsc ...
Now seriously, KX_KetsjiEngine::RenderFonts() could be moved to inside the KX_Scene class. It probably should (so I could call it from inside KX_Dome::RenderDomeFrame()). Not critical, so not changing it for now.
Initially problem was caused by updated version of mingw-runtime which
changed behavior of snprintf and vsnprintf so %lf isn't anymore valid
for doubles.
According to manpages, %f is a correct format for snprintf for doubles.
- change short -> char for flags that support it.
- add pose 'temp' pointer to use for outliner drawing (was using 'prev' and restoring which seems dodjy)
* Added an optional array of MDeformModifier pointers, to avoid another search based on defgrp_idx.
* Split out "add/remove verts from vgroup" functions, preparing their move to deform.c (if their current form is validated!).
Also fixed a bug: when another set mode than "All Vertices" was used and resulting set of verts was empty, all vertices was used, instead of just returning org, unmodified data!
Use this to raise errors when assigning a string property fails even though the value to assign *is* a string.
Before:
TypeError: bpy_struct: item.attr= val: Object.name expected a string type, not str
After:
TypeError: bpy_struct: item.attr= val: Object.name error assigning string, UnicodeEncodeError('utf-8' codec can't encode character '\udce9' in position 23: surrogates not allowed)
