Blender Git Commits

OpenGL: don't poll for errors, rely on KHR_debug

Errors are caught & reported by our GL debug callback. This gives us way more useful information than sporadic calls to glGetError.

I removed almost all use of glGetError, including our own GPU_ASSERT_NO_GL_ERRORS and GPU_CHECK_ERRORS_AROUND macros.

Still used in rna_Image_gl_load because it passes unvalidated input to OpenGL functions.

Still used in gpu_state_print_fl_ex as an exception handling hack -- will rewrite this soon.

The optimism embodied by this commit will not prevent OpenGL errors. We need to analyze what would cause GL to fail at certain points and proactively intercept these failures. Or guarantee they can't happen.

Alembic: fix crash accessing invalid objects.

Fix T49111: Automatically add file path suffix for Alembic and Collada
export.

This aligns the behaviour of the file selection with the other
exporters. The Alembic case would fail if the filepath did not have an
extension set.

Also set a default file name for the Alembic export operator in case the
Blender file was not saved before exporting.

Cleanup: Remove redundant comment

Cycles: Fix wrong allocator used for spatial builder

Cleanup: some bad sizeof() usages.

Cleanup/security fix: do not use strcpy (at least in new code).

This function is only really secure in a very limited amount of cases,
and can especially bite you later if you change some buffer sizes...
So not worth bothering with it, just always use BLI_strncpy instead.

Cleanup for previous commit (nasty IDE replacing tads with spaces, tsst)

Sorry for the noise :|

Two fixes for optional ndof & fix bplayer for that...

Seriuosly, guys... Please always check that kind of changes with and without affected option, and in full build case.

Freestyle: fix wrong arg order, and cleanup confusing loop (both reported by coverity).

Error: `origin` and `edge` args were swapped in final `FindOccludee()` call of `ViewMapBuilder::ComputeRayCastingVisibility()`

Cleanup: main for loop in `Strip::createStrip()` was really confusing (though correct),
generated a false positive in coverity scan, now should be cleaner how it loops over its vprev/v/v2
triplet of consecutive items.

Final UI messages fixes (for this session...).

Fix bplayer (c) after NDOF changes from merwin

Cleanup: ifdef function which is only used from ifdef-ed code

NDOF: compile 3D mouse code only if WITH_INPUT_NDOF

When WITH_INPUT_NDOF is disabled, 3D mouse handling code is removed
from:

- GHOST (was mostly done, finished the job)
- window manager
- various editors
- RNA
- keymaps

The input tab of user prefs does not show 3D mouse settings. Key map
editor does not show NDOF mappings.

DNA does not change.

On my Mac the compiled binary is 42KB smaller after this change. It
runs fine WITH_INPUT_NDOF on or off.

NDOF: suppress buttons debug log

Accidentally left this in the 3D mouse code. Mac only.

Fix compiler warning after fix for T48913.

Code cleanup to use array.data() rather than &array[0]

(Now without the build errors)

Revert "Code cleanup to use array.data() rather than &array[0]."

This reverts commit 40b367479c6fe23d6f2b6d822f2d5266485619f3.

Didn't build or solve any known issue. Please don't push changes without
testing them first.

Code cleanup to use array.data() rather than &array[0].

These latter can cause MSVC debug asserts if the array is empty. With C++11
we'll be able to do this for std::vector later. This hopefully fixes an assert
in the Cycles subdivision code.

Fix T48913: cycles viewport render stuck in loop due to non-unique dupli ID.