[Bf-committers] [22334] branches/blender2.5/blender: fix for build problem with audiospace and implicit declaration.

[Campbell Barton]

Agree its a bit of a mess, however my reason for referencing ./intern
was that other areas of the code should not have to have includes for
both audiospace and audiospace/intern
Looks like this happens all over so Im not sure whats best option in this case.

On Mon, Aug 10, 2009 at 11:37 AM, GSR<gsr.b3d at infernal-iceberg.com> wrote:
> Hi,
> nicholasbishop at gmail.com (2009-08-10 at 1418.34 -0400):
>> I agree with his change; it seems like a bad idea to change all the
>> build references for audio to include a reference to the *internal*
>> header directory for audaspace. If the directory structure is at some
>> point changed, it's better to have it just change in the external
>> header (AUD_C-API.h). Also, the hardcoded path is still relative, so I
>> don't see how it's any more hardcoded either way, you're still
>> explicitly referencing a file.
>
> The current tree is a chaos, you can find this:
>
> intern/boolop/intern/BOP_Interface.cpp:31:#include "../../bsp/intern/BSP_CSGMesh_CFIterator.h"
> intern/boolop/extern/BOP_Interface.h:31:#include "../../bsp/intern/BSP_CSGMesh.h"
> intern/iksolver/test/ik_glut_test/intern/ChainDrawer.h:37:#     include "../intern/IK_QChain.h"
> intern/iksolver/test/ik_glut_test/intern/ChainDrawer.h:38:#     include "../intern/IK_QSolver_Class.h"
> intern/ghost/intern/GHOST_C-api.cpp:46:#include "intern/GHOST_Debug.h"
> intern/ghost/intern/GHOST_C-api.cpp:51:#include "intern/GHOST_CallbackEventConsumer.h"
> source/blender/render/intern/source/pipeline.c:65:#include "intern/openexr/openexr_multi.h"
> source/blender/blenkernel/intern/node.c:71:#include "intern/CMP_util.h" /* stupid include path... */
> source/blender/blenkernel/intern/node.c:75:#include "intern/TEX_util.h"
> source/blender/blenkernel/intern/image.c:47:#include "intern/openexr/openexr_multi.h"
>
> Or other places where all the headers are at the same level, and thus
> installing them works (via the *.h glob) even if they are not, for example:
>
> ./intern/memutil/MEM_Allocator.h
> ./intern/memutil/MEM_CacheLimiter.h
> ./intern/memutil/MEM_CacheLimiterC-Api.h
> ./intern/memutil/MEM_NonCopyable.h
> ./intern/memutil/MEM_RefCountPtr.h
> ./intern/memutil/MEM_RefCounted.h
> ./intern/memutil/MEM_RefCountedC-Api.h
> ./intern/memutil/MEM_SmartPtr.h
>
>> It's not a huge deal either way, but I prefer not to alter a bunch of
>> CMakeLists unnecessarily, so some consensus would be good :)
>
> Yep, and then I could fix makefiles in one or the other direction,
> because the source tree has examples for everything, with intern or
> without intern, with hardcoded -I or ones that depend on variables. If
> the way to go is to use in tree headers, then we can save a lot of
> install operations.
>
> GSR
>
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at blender.org
> http://lists.blender.org/mailman/listinfo/bf-committers
>



-- 
- Campbell