Re: [osg-users] Problem compiling OSG from SVN in Cygwin

2007-12-20 Thread Robert Osfield
Hi Brian,

I can't spot what might be up, there wasn't any changes to the core
OSG to support the cfg plugin, so the addition of it should just the
building of it and nothing else.   Other changes checked in for other
reasons might affect the build and this in turn could effect the cfg
plugin.   Its not entirely clear to me which bits work out of the box
with the SVN version of the OSG, so its not clear what needs to be
done to fix the build.

Working on the assumption that its just the include order in the cfg
plugin that is leading to some functions/macros being
redefined/redefined I've moved the OSG includes to the tops of the
files - things still compile under Linux so it would seem the changes
are benign... but of course can't say how Cygwin will fare with them,
perhaps no effect, perhaps a fix, perhaps make it worse...

Could you check out the latest version in SVN and let me know exactly
what things work/break out of the box without any further changes.

Cheers,
Robert.

On Dec 19, 2007 6:51 PM, Brian Keener [EMAIL PROTECTED] wrote:
 Robert Osfield wrote:
  Couple of observations, include/osg/Math hasn't changed for a very
  long time, so it suggest that these methods aren't compiled by the
  compile unless one compiles the .cfg plugin

 No actually - if I change line 206 or line 207 to something like
 _isnanf and _isnand (1 underscore) then very early in the build it
 breaks on Animationpath.h and .cpp and if I change to lines back to the
 svn version then Animationpath.h and .cpp build as they should.

  , Which does perplex me as
  I'm sure these methods are used elsewhere, so what could be up?

 Agreed.  See my response above as that is what I noticed - certain
 changes to fix the .cfg plugin broke Animationpath.h and .cpp in the
 build very early but my changes to line 206 and 207 using isnanf and
 __isnand (2 underscores) resulted in everything building as they
 should.  If I look at /usr/include/math.h in Cygwin it defines isnanf
 and isnand.

  I don't know.  Perhaps one could look at the include order to see in
  .cpp/.h files in the .cfg plugin are doing anything unusual.

 That was my thought I starting looking to see what was included or not
 included or the order they were included in Animationpath.h and .cpp vs
 the .cfg plugin but nothing yet.

 Thanks.


 bk



 ___
 osg-users mailing list
 osg-users@lists.openscenegraph.org
 http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org

___
osg-users mailing list
osg-users@lists.openscenegraph.org
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org


Re: [osg-users] Problem compiling OSG from SVN in Cygwin

2007-12-19 Thread Brian Keener
Robert Osfield wrote:
 Couple of observations, include/osg/Math hasn't changed for a very
 long time, so it suggest that these methods aren't compiled by the
 compile unless one compiles the .cfg plugin

No actually - if I change line 206 or line 207 to something like 
_isnanf and _isnand (1 underscore) then very early in the build it 
breaks on Animationpath.h and .cpp and if I change to lines back to the 
svn version then Animationpath.h and .cpp build as they should.  

 , Which does perplex me as
 I'm sure these methods are used elsewhere, so what could be up?

Agreed.  See my response above as that is what I noticed - certain 
changes to fix the .cfg plugin broke Animationpath.h and .cpp in the 
build very early but my changes to line 206 and 207 using isnanf and 
__isnand (2 underscores) resulted in everything building as they 
should.  If I look at /usr/include/math.h in Cygwin it defines isnanf 
and isnand.

 I don't know.  Perhaps one could look at the include order to see in
 .cpp/.h files in the .cfg plugin are doing anything unusual.

That was my thought I starting looking to see what was included or not 
included or the order they were included in Animationpath.h and .cpp vs 
the .cfg plugin but nothing yet.

Thanks.

bk



___
osg-users mailing list
osg-users@lists.openscenegraph.org
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org


[osg-users] Problem compiling OSG from SVN in Cygwin

2007-12-18 Thread Brian Keener
Hello Robert,

It's your worst nightmare again.  I tried compiling the svn version 
again from trunk since you were asking folks to check various 
platforms.

I downloaded fresh from svn and did the ccmake and then ran make and I 
get this error:

[ 78%] Building CXX object 
src/osgPlugins/ive/CMakeFiles/osgdb_ive.dir/Visibilit
yGroup.o
Linking CXX shared module 
./../../lib/osgPlugins-2.2.0/cygwin_osgdb_ive.dll
[ 78%] Built target osgdb_ive
Scanning dependencies of target osgdb_cfg
[ 78%] Building CXX object 
src/osgPlugins/cfg/CMakeFiles/osgdb_cfg.dir/CameraCon
fig.o
In file included from /usr/src/OpenSceneGraph/include/osg/Vec2f:17,
 from /usr/src/OpenSceneGraph/include/osg/Vec2d:17,
 from /usr/src/OpenSceneGraph/include/osg/Vec3d:17,
 from /usr/src/OpenSceneGraph/include/osg/Matrixd:18,
 from /usr/src/OpenSceneGraph/include/osg/Matrix:17,
 from 
/usr/src/OpenSceneGraph/src/osgPlugins/cfg/Camera.h:19,
 from 
/usr/src/OpenSceneGraph/src/osgPlugins/cfg/CameraConfig.h:
19,
 from 
/usr/src/OpenSceneGraph/src/osgPlugins/cfg/CameraConfig.cp
p:34:
/usr/src/OpenSceneGraph/include/osg/Math: In function `bool 
osg::isNaN(float)':
/usr/src/OpenSceneGraph/include/osg/Math:206: error: `isnan' undeclared 
(first u
se this function)
/usr/src/OpenSceneGraph/include/osg/Math:206: error: (Each undeclared 
identifier
 is reported only once for each function it appears in.)
/usr/src/OpenSceneGraph/include/osg/Math: In function `bool 
osg::isNaN(double)':

/usr/src/OpenSceneGraph/include/osg/Math:207: error: `isnan' undeclared 
(first u
se this function)
make[2]: *** 
[src/osgPlugins/cfg/CMakeFiles/osgdb_cfg.dir/CameraConfig.o] Error
1
make[1]: *** [src/osgPlugins/cfg/CMakeFiles/osgdb_cfg.dir/all] Error 2
make: *** [all] Error 2

[EMAIL PROTECTED] /usr/develop/obj/osg



So then in include/osg/Math I change line 206 from this:

inline bool isNaN(float v) { return isnan(v); }

to 

inline bool isNaN(float v) { return isnanf(v); }

and I change line 207 to return various forms of isnand, _isnan, 
_isnand just to see what will happen and 
then the build blows up much earlier on Animationpath in OpenThreads as 
follows:

[EMAIL PROTECTED] /usr/develop/obj/osg
$ make
[  0%] Built target OpenThreads
[  1%] Building CXX object src/osg/CMakeFiles/osg.dir/AnimationPath.o
In file included from /usr/src/OpenSceneGraph/include/osg/Vec2f:17,
 from /usr/src/OpenSceneGraph/include/osg/Vec2d:17,
 from /usr/src/OpenSceneGraph/include/osg/Vec3d:17,
 from /usr/src/OpenSceneGraph/include/osg/Matrixf:18,
 from 
/usr/src/OpenSceneGraph/include/osg/AnimationPath:21,
 from 
/usr/src/OpenSceneGraph/src/osg/AnimationPath.cpp:13:
/usr/src/OpenSceneGraph/include/osg/Math: In function `bool 
osg::isNaN(double)':

/usr/src/OpenSceneGraph/include/osg/Math:207: error: `_isnand' 
undeclared (first
 use this function)
/usr/src/OpenSceneGraph/include/osg/Math:207: error: (Each undeclared 
identifier
 is reported only once for each function it appears in.)
make[2]: *** [src/osg/CMakeFiles/osg.dir/AnimationPath.o] Error 1
make[1]: *** [src/osg/CMakeFiles/osg.dir/all] Error 2
make: *** [all] Error 2

[EMAIL PROTECTED] /usr/develop/obj/osg
$

This error which would lead me to believe that during the early stages 
of the compile - isnan is defined and then it disappears by the time I 
get to osgdb_config.

Further on this is that I found if I change line 206 to 

inline bool isNaN(float v) { return isnanf(v); }

and line 207 to 

inline bool isNaN(double v) { return __isnand(v); }

then everything compile - any thought on this

bk




___
osg-users mailing list
osg-users@lists.openscenegraph.org
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org