[Bf-committers] Fedora package patch cleanup.

[Campbell Barton]

Hi Richard,

Reply inline...

On Wed, Sep 7, 2011 at 4:11 AM, Richard Shaw <hobbes1069 at gmail.com> wrote:
> Hey guys, I inhereted a lot of patches when I started trying to build
> blender 2.5X for Fedora and I've weeded out a lot of them but there's
> a couple still around that I'd like to remove if I can.
>
> 1. This does something with bmake. I'm using cmake for building, do I need this?
>
> --- blender-2.44/release/plugins/bmake.bid      2004-12-27
> 20:28:48.000000000 +0100
> +++ blender-2.44/release/plugins/bmake  2007-08-09 18:13:34.000000000 +0200
> @@ -60,10 +60,10 @@ elif ( test $UNAME = "OpenBSD" ) then
>        LDFLAGS="-Bshareable";
>  elif ( test $UNAME = "Linux" ) then
>        CC="gcc";
> -       CFLAGS="-fPIC -funsigned-char -O2";
> +       CFLAGS="${RPM_OPT_FLAGS} -fPIC -funsigned-char";
>
> -       LD="ld";
> -       LDFLAGS="-Bshareable";
> +       LD="gcc";
> +       LDFLAGS="--shared ${RPM_OPT_FLAGS}";
>  elif ( test $UNAME = "SunOS" ) then
>        CC="cc";
>        CFLAGS="-O";
>
> 2. I'm assuming this has something to do with 64bit system builds?

has anyone even tested this with 2.5x?
I think if we bring plugins back properly we'll use cmake/scons.
Patch might be find but not good to apply if we don't test.

I wouldn't worry about this - its pretty much deprecated.

> Index: source/gameengine/Expressions/Value.cpp
> ===================================================================
> --- source/gameengine/Expressions/Value.cpp.orig        2010-04-18
> 12:28:37.000000000 +0200
> +++ source/gameengine/Expressions/Value.cpp     2010-10-28
> 14:56:24.000000000 +0200
> @@ -543,8 +543,8 @@ CValue* CValue::ConvertPythonToValue(PyO
>                CListValue* listval = new CListValue();
>                bool error = false;
>
> -               int i;
> -               int numitems = PyList_Size(pyobj);
> +               Py_ssize_t i;
> +               Py_ssize_t numitems = PyList_Size(pyobj);
>                for (i=0;i<numitems;i++)
>                {
>                        PyObject* listitem = PyList_GetItem(pyobj,i);
> /* borrowed ref */
>
> 3. I don't see how this does anything useful, but I may have already
> removed segments that no longer applied cleanly:

Applied changes to source/gameengine/Expressions/Value.cpp (this block
is actually #if 0'd but using Py_ssize_t for list size is reasonable
since thats what python uses internally).

> Index: release/plugins/sequence/dnr.c
> ===================================================================
> --- release/plugins/sequence/dnr.c.orig 2008-11-12 23:16:53.000000000 +0200
> +++ release/plugins/sequence/dnr.c      2010-10-28 14:56:30.000000000 +0200
> @@ -110,9 +110,12 @@ static void doit(unsigned char * src_, u
>        unsigned char * dst = dst_;
>
>        while (count--) {
> -               *dst = table[(*src++ << 8) | *dst]; dst++;
> -               *dst = table[(*src++ << 8) | *dst]; dst++;
> -               *dst = table[(*src++ << 8) | *dst]; dst++;
> +               *dst = table[(*src++ << 8) | *dst];
> +               dst++;
> +               *dst = table[(*src++ << 8) | *dst];
> +               dst++;
> +               *dst = table[(*src++ << 8) | *dst];
> +               dst++;
>                *dst++ = *src++;
>
> 4. I don't know anything about this one...

Just a style change, don't see the point in downstream maintaining
style edits, suggest you ignore this one.

> Index: extern/bullet2/src/LinearMath/btVector3.h
> ===================================================================
> --- extern/bullet2/src/LinearMath/btVector3.h.orig      2009-03-17
> 23:44:58.000000000 +0200
> +++ extern/bullet2/src/LinearMath/btVector3.h   2010-10-28
> 14:56:38.000000000 +0200
> @@ -63,7 +63,8 @@ public:
>        public:
>
>   /**@brief No initialization constructor */
> -       SIMD_FORCE_INLINE btVector3() {}
> +       SIMD_FORCE_INLINE btVector3() {
> +       m_floats[0]=m_floats[1]=m_floats[2]=m_floats[3]=btScalar(0.);}
>
>
>
> Index: extern/bullet2/src/BulletCollision/BroadphaseCollision/btQuantizedBvh.h
> ===================================================================
> --- extern/bullet2/src/BulletCollision/BroadphaseCollision/btQuantizedBvh.h.orig
>       2009-09-14 16:45:32.000000000 +0200
> +++ extern/bullet2/src/BulletCollision/BroadphaseCollision/btQuantizedBvh.h
>    2010-10-28 14:56:38.000000000 +0200
> @@ -94,7 +94,8 @@ ATTRIBUTE_ALIGNED16 (struct) btOptimized
>
>        //for child nodes
>        int     m_subPart;
>        int     m_triangleIndex;
> -       int     m_padding[5];//bad, due to alignment
> +       int:5*8*sizeof(int);
> +//     int     m_padding[5];//bad, due to alignment
>
>
>  };

Best submit these to the bullet project directly, we only apply fixes
in blender's bullet if they fix crashes or bad bugs, then try push
upstream, but don't think this is important enough to be applied to
blender.

> Thanks,
> Richard
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at blender.org
> http://lists.blender.org/mailman/listinfo/bf-committers
>



-- 
- Campbell