Blender Git Loki

Rewired python for OSX. The Project Builder project now uses the Python that
comes with OSX in /usr/local.
Removed some frozen python files that were not generated by the freeze script.
mxTextTools is still disabled because of a link error.
Maarten

Rewired python for OSX. The Makefiles now use the Python that comes with OSX
in /usr/local.
mxTextTools is still disabled because of a link error.
Maarten

added the DELKEY stuff...
else if(dev==DELKEY) {
if(but->pos>=0 && but->pos<strlen(str)) {
for(x=but->pos; x<=strlen(str); x++)
str[x]= str[x+1];
str[--len]=' ';
dodraw= 1;
}
}

Kent

Fix so that rgba targa files view correctly in gimp:
Submitted to bf-committers by Chris Want

Kent

Index: targa.c
===================================================================
RCS file: /cvs01/blender/source/blender/imbuf/intern/targa.c,v
retrieving revision 1.4
diff -u -r1.4 targa.c
--- targa.c 2002/12/20 01:29:14 1.4
+++ targa.c 2002/12/21 09:50:24
@@ -289,6 +289,12 @@

if (flags & IB_ttob) buf[17] ^= 0x20;

+ /* Don't forget to indicate that your 32 bit
+ * targa uses 8 bits for the alpha channel! */
+ if (ibuf->depth==32) {
+ buf[17] |= 0x08;
+ }
+
if (write(file, buf, 18) != 18) return (0);
if (ibuf->cmap){
for (i = 0 ; i<ibuf->maxcol ; i++){

This is sort of a two parter from the mailling list but really its the
same thing:

Added the xpcom include and added a flag to use mozilla vs netscape.
Which basically adds some ifdefs to
_Blender3DPlugin_implementation_.cpp

Basically I modified the nan_definitions.mk in the following way
(You'll need to update yours don't forget) ;)
Also updated the makefiles to use the new vars and flushed them out
so they first include whatever nspr is set to and then look in
NAN_MOZILLA_INC

I wasn't sure what the NAN_MOZILLA_INC and _LIB defaults should be
since were assuming they point to a netscape dir by default (or maybe
old mozilla???)
maybe they should be
export NAN_MOZILLA_INC ?= $(LCGDIR)/netscape/include
export NAN_MOZILLA_LIB ?= $(LCGDIR)/netscape/lib/

instead...

Anyone have any opinions on the defaults?

Kent
--
mein@cs.umn.edu

Index: nan_definitions.mk
===================================================================
RCS file: /cvs01/blender/source/nan_definitions.mk,v
retrieving revision 1.8
diff -u -r1.8 nan_definitions.mk
--- nan_definitions.mk 2002/12/19 21:12:58 1.8
+++ nan_definitions.mk 2002/12/20 19:28:51
@@ -75,7 +75,13 @@
export NAN_TEST_VERBOSITY ?= 1
export NAN_ZLIB ?= $(LCGDIR)/zlib
export NAN_BMFONT ?= $(LCGDIR)/bmfont
- export NAN_MOZILLA ?= $(LCGDIR)/mozilla
+ # Uncomment the following line to use Mozilla inplace of netscape
+ # CPPFLAGS +=-DMOZ_NOT_NET
+ # Location of MOZILLA/Netscape header files...
+ export NAN_MOZILLA_INC ?= $(LCGDIR)/mozilla/include
+ export NAN_MOZILLA_LIB ?= $(LCGDIR)/mozilla/lib/
+ # Will fall back to look in NAN_MOZILLA_INC/nspr and NAN_MOZILLA_LIB
+ # if this is not set.
export NAN_NSPR ?= $(LCGDIR)/nspr

Shrunk the hack so it didn't get used in newer versions of OpenAL which
seems to be working.

Kent

(diff follows)

Index: SND_OpenALDevice.cpp
===================================================================
RCS file: /cvs01/blender/source/gameengine/SoundSystem/openal/SND_OpenALDevice.cpp,v
retrieving revision 1.4
diff -u -r1.4 SND_OpenALDevice.cpp
--- SND_OpenALDevice.cpp 2002/11/25 15:29:57 1.4
+++ SND_OpenALDevice.cpp 2002/12/20 19:26:04
@@ -59,7 +59,7 @@
#include <fcntl.h>

/* untill openal gets unified we need this hack for non-windows systems */
-#ifndef WIN32
+#if !defined(WIN32) && !defined(ALC_MAJOR_VERSION)

Added a link for where to get glut

Kent
--
mein@cs.umn.edu

Fixed up the header comment so it didn't produce a warning about
comment inside of a comment

Also removed the config.h thing since this is a .h

Hopefully I'll get around to the other .h's soon.

Kent
--
mein@cs.umn.edu

Back at it ;)

Removed an unused var...
Here is the cvs diff.

Kent

Index: bsp/intern/BSP_CSGUserData.cpp
===================================================================
RCS file: /cvs01/blender/intern/bsp/intern/BSP_CSGUserData.cpp,v
retrieving revision 1.4
diff -u -r1.4 BSP_CSGUserData.cpp
--- BSP_CSGUserData.cpp 2002/11/25 09:52:48 1.4
+++ BSP_CSGUserData.cpp 2002/12/20 09:37:58
@@ -79,7 +79,6 @@
int record_index
){
if (m_width) {
- int output = Size();
IncSize();

Ok last change for the night :)
Time to go watch er.

Note this was already done with face_old, img_old was probably just
overlooked.
(If you guys didn't notice I was sick for a couple of days but I'm back now
as you can tell)

Kent
--
mein@cs.umn.edu

Index: editface.c
===================================================================
RCS file: /cvs01/blender/source/blender/src/editface.c,v
retrieving revision 1.3
diff -u -r1.3 editface.c
--- editface.c 2002/11/25 12:02:05 1.3
+++ editface.c 2002/12/20 02:50:55
@@ -1202,7 +1202,7 @@
TFace *face, *face_old = 0;
short xy[2], xy_old[2];
//int a, index;
- Image *img, *img_old;
+ Image *img, *img_old = 0;

In stubs.c I added a dummy return to functions that were non void types.

in readfile.c I renamed some localvars that were named main to mainl in
a couple of functions.

in action.c I initalized a var to 0. This is probably not needed but
its a little safer incase someone adds more modes...
I've included a diff of this last one below.

Kent

Index: action.c
===================================================================
RCS file: /cvs01/blender/source/blender/blenkernel/intern/action.c,v
retrieving revision 1.6
diff -u -r1.6 action.c
--- action.c 2002/11/25 12:01:52 1.6
+++ action.c 2002/12/20 02:06:27
@@ -153,7 +153,7 @@
){
bConstraint *dcon;
const bConstraint *scon;
- float dstweight;
+ float dstweight = 0;

switch (mode){
case POSE_BLEND:

Two more variables that were not initalized.

cvs diff included below.

Kent
--
mein@cs.umn.edu

Index: png_encode.c
===================================================================
RCS file: /cvs01/blender/source/blender/imbuf/intern/png_encode.c,v
retrieving revision 1.3
diff -u -r1.3 png_encode.c
--- png_encode.c 2002/11/25 12:02:00 1.3
+++ png_encode.c 2002/12/20 01:28:18
@@ -91,7 +91,7 @@
unsigned char *pixels = 0;
unsigned char *from, *to;
png_bytepp row_pointers = 0;
- int i, bytesperpixel, color_type;
+ int i, bytesperpixel, color_type = PNG_COLOR_TYPE_GRAY;
FILE *fp = 0;

Index: targa.c
===================================================================
RCS file: /cvs01/blender/source/blender/imbuf/intern/targa.c,v
retrieving revision 1.3
diff -u -r1.3 targa.c
--- targa.c 2002/11/25 12:02:00 1.3
+++ targa.c 2002/12/20 01:28:20
@@ -247,7 +247,7 @@
char buf[20];
FILE *fildes;
int i;
- short ok;
+ short ok = 0;

Ok since no one cares about this I put a 5 second fix on it and won't
worry about it anymore :)

(initalized it to 0 and then later return 0 if its still 0 before doing
the work)

Kent
--
mein@cs.umn.edu

Fixed two uninitalized vars:

Kent

/cvs01/blender/source/blender/imbuf/intern/iris.c,v
retrieving revision 1.3
diff -u -r1.3 iris.c
--- iris.c 2002/11/25 12:02:00 1.3
+++ iris.c 2002/12/19 22:12:53
@@ -212,7 +212,7 @@
/* unsigned int *tab; */
/* int len; */
{
- int r;
+ int r = 0;

while(len) {
r = putlong(outf,*tab++);
@@ -548,7 +548,7 @@
{
FILE *outf;
IMAGE *image;
- int tablen, y, z, pos, len;
+ int tablen, y, z, pos, len = 0;
int *starttab, *lengthtab;
unsigned char *rlebuf;

Added the the magic compiler flag that will turn on game-engine integration with Blender. The line is commented by default as was discussed on the list.
Maarten

I initalized mask to 0 in IMB_converttocmap

This may not be correct but at least now its predictable.

Kent

Used NAN_ODE to point to the ode lib location.
Maarten

Overrided NAN_ODE in OSX to point to the precompiled ode lib in the lib tree.
Maarten

-static void SleepTillEvent(Display *display, GHOST_TUns64 maxSleep) {
+static void SleepTillEvent(Display *display, GHOST_TInt64 maxSleep) {

maxSleep is checked against -1 so it needs to have a sign.
This could even fix some stability problems since -1 is passed in sometimes
in the code.

Kent
--
mein@cs.umn.edu

Fixed this warning:

IK_QSegment.h: In constructor `IK_QSegment::IK_QSegment(MT_Point3,
MT_Matrix3x3, double, MT_ExpMap)':
IK_QSegment.h:271: warning: member initializers for `MT_Scalar
IK_QSegment::m_length'
IK_QSegment.h:270: warning: and `MT_ExpMap IK_QSegment::m_q'
IK_QSegment.cpp:56: warning: will be re-ordered to match declaration order
IK_QSegment.h: In constructor `IK_QSegment::IK_QSegment()':
IK_QSegment.h:271: warning: member initializers for `MT_Scalar
IK_QSegment::m_length'
IK_QSegment.h:270: warning: and `MT_ExpMap IK_QSegment::m_q'
IK_QSegment.cpp:73: warning: will be re-ordered to match declaration order

Kind of goofy its just the order of which 2 private vars are defined.

Kent
--
mein@cs.umn.edu