openvrml-develop — OpenVRML developers mailing list

[openvrml-develop] openvrml-0.16.6 binaries now available for Mac, Windows

[Philip L. <ph...@ed...>]

Hi all,

In preparation for release of a new version of ARToolKit, which  
includes a renderer based on openvrml, I am making available binary  
builds of openvrml-0.16.6 for Mac OS X and Windows 32-bit.

The packages are now available for download from ARToolworks, Inc.:
	http://www.artoolworks.com/dist/openvrml/0.16.6/

The Mac build is packaged as a standalone set of headers and dynamic  
libraries, suitable for installation into an application bundle. This  
is in contrast to the openvrml-0.16.6 package available through the  
Fink Mac package manager, which builds openvrml as system-shared  
libraries.

The Windows build is suitable for users building against openvrml  
using Microsoft Visual Studio 2005. It does not contain an installer  
at present.

Please direct any support queries to this list, not to ARToolworks  
staff.

Regards,

Phil.

Philip Lamb
CTO, ARToolworks, Inc.
http://www.artoolworks.com/

Re: [openvrml-develop] OpenVRML 0.16.6 on Windows.. still no success.

[Braden M. <br...@en...>]

On Wed, 2007-10-24 at 11:21 +1300, Philip Lamb wrote:
> On 23/10/2007, at 5:33 PM, Braden McDaniel wrote:

[snip]

> > It's been a while since I set up libpng and libjpeg to build with on
> > Windows. I'm guilty of punting here. Can you tell me more about the C
> > runtime incompatibilities?
> 
> Sure.. one or other of libpng or libjpeg from gnuwin32 is built  
> against Visual Studio 2003's DLL runtime (MSVCRT.dll). In Visual  
> Studio 2005 there is no static C runtime model (all C/C++ code links  
> MSVCRT.dll/MSVCPRT.dll). So the externals in libpng.lib or  
> libjpeg.lib must be satisfied from Visual Studio 2005's version of  
> the DLL runtimes. As this version is a few symbols short (_iob, and  
> some from the STL) this results in missing symbols.  You cannot just  
> supply MSVCRT71.dll, because of Microsoft's joyous DLL-hell hack  
> (Windows side-by-side assemblies). Nor is there a stub library one  
> can link to for the missing symbols.

Lovely. Did you end up building from the official libjpeg and libpng
distributions? Or something modified to be more Windows-friendly?

> > I'll Feel The Pain and get this set up.
> 
> If in fact you have done a build of OpenVRML on Windows after each  
> release to test it, it would probably be very helpful to just zip  
> that build up and post it as a binary release.

Because I think that supporting binaries without an actual installer
would be more difficult than helping people build the source.

> >> 3) try to compile openvrml. Fix issues with openvrml.vsproj file,
> >> including include paths picking up linux openvrml-config instead of
> >> windows openvrml config.
> >
> > That's not a problem with the default include paths. It's a problem  
> > with
> > putting platform-specific build files in a source tree that you're
> > trying to use with multiple platforms. In other words, it's a problem
> > with building in the srcdir. Don't Do That.
> >
> 
> This was one I asked about earlier. I think that in the vsproj  
> release build, ..\..\..\..\..\src\libopenvrml is ahead of "." on the  
> "additional includes" line,

"." is first in the list for both Debug and Release configurations in
the project files that are included with 0.16.6.

>  which makes it pick up the Linux openvrml- 
> config instead of the MSVS one. Its OK on the debug build settings.

It won't do that if that file doesn't exist--and it doesn't if you don't
build in the srcdir.

> Another gotcha in this area which I uncovered yesterday.. the latest  
> DirectX SDK's (don't ask why I have this installed) put themselves at  
> the head of Visual Studio's global include path, and they include a  
> version of strsafe.h which is incompatible with the version in the  
> Platform SDK. The error caused by this is a missing symbol  
> STRSAFE_LPCSTR in compiling vrml97node.cpp. The solution is to move  
> the DirectX SDK include path to the tail of the includes list so that  
> #include <strsafe.h> picks up strsafe from the Platform SDK.

Nice.

> > Can you please file a bug in SourceForge and attach the .wrl file  
> > you're
> > using? I will take a look at this.
> 
> Done.

Thank you.

-- 
Braden McDaniel                           e-mail: <br...@en...>
<http://endoframe.com>                    Jabber: <br...@ja...>

Re: [openvrml-develop] OpenVRML 0.16.6 on Windows.. still no success.

[Philip L. <ph...@ed...>]

On 23/10/2007, at 5:33 PM, Braden McDaniel wrote:
>
> In general, the warnings are there because they are silly and because
> disabling them doesn't really work. If there are warnings you  
> believe to
> be pointing to legitimate issues, it is possible you have found
> something that I overlooked in the sea of worthless spew (and  
> something
> that gcc does not complain about).

Well, I'll admit I didn't wade through every one, and I also observed  
that disabling them doesn't work. Ah, Microsoft's quality development  
tools.


>
>> Here are the steps so far:
>> 1) try to compile boost on Windows. Give up. Use boost-
>> consulting.com's somewhat broken boost binaries installer. OK.
>> 2) try to compile openvrml against gnuwin32's freetype/zlib/libpng/
>> jpeg. C-runtime model incompatibility forces download and build zlib,
>> jpeg, libpng, freetype from source. Oh joy.
>
> It's been a while since I set up libpng and libjpeg to build with on
> Windows. I'm guilty of punting here. Can you tell me more about the C
> runtime incompatibilities?

Sure.. one or other of libpng or libjpeg from gnuwin32 is built  
against Visual Studio 2003's DLL runtime (MSVCRT.dll). In Visual  
Studio 2005 there is no static C runtime model (all C/C++ code links  
MSVCRT.dll/MSVCPRT.dll). So the externals in libpng.lib or  
libjpeg.lib must be satisfied from Visual Studio 2005's version of  
the DLL runtimes. As this version is a few symbols short (_iob, and  
some from the STL) this results in missing symbols.  You cannot just  
supply MSVCRT71.dll, because of Microsoft's joyous DLL-hell hack  
(Windows side-by-side assemblies). Nor is there a stub library one  
can link to for the missing symbols.

>
> I'll Feel The Pain and get this set up.

If in fact you have done a build of OpenVRML on Windows after each  
release to test it, it would probably be very helpful to just zip  
that build up and post it as a binary release.

>> 3) try to compile openvrml. Fix issues with openvrml.vsproj file,
>> including include paths picking up linux openvrml-config instead of
>> windows openvrml config.
>
> That's not a problem with the default include paths. It's a problem  
> with
> putting platform-specific build files in a source tree that you're
> trying to use with multiple platforms. In other words, it's a problem
> with building in the srcdir. Don't Do That.
>

This was one I asked about earlier. I think that in the vsproj  
release build, ..\..\..\..\..\src\libopenvrml is ahead of "." on the  
"additional includes" line, which makes it pick up the Linux openvrml- 
config instead of the MSVS one. Its OK on the debug build settings.

Another gotcha in this area which I uncovered yesterday.. the latest  
DirectX SDK's (don't ask why I have this installed) put themselves at  
the head of Visual Studio's global include path, and they include a  
version of strsafe.h which is incompatible with the version in the  
Platform SDK. The error caused by this is a missing symbol  
STRSAFE_LPCSTR in compiling vrml97node.cpp. The solution is to move  
the DirectX SDK include path to the tail of the includes list so that  
#include <strsafe.h> picks up strsafe from the Platform SDK.

> Can you please file a bug in SourceForge and attach the .wrl file  
> you're
> using? I will take a look at this.

Done. I've attached 2 zip archives to the bug. One (sample wrl.zip)  
contains two vrml files, one (apple.wrl) is a simple mesh with jpeg  
texture, the other (snoman.wrl) has no textures, but has some  
animation. Both load and render OK (they do dump a few warnings about  
non-normalized rotation vectors) using sdl-viewer (on Mac OS X using  
openvrml-sdl-viewer from my openvrml-016 package from fink). The  
second zip archive contains the VS project files I used to build  
openvrml (a diff with the ones in openvrml-0.16.6.tar.gz should  
quickly show the changes).

I've also packaged my builds of freetype, zlib, libpng, and jpeg into  
a package, now available at http://www.eden.net.nz/7/20071024/ 
zlib_freetype231_jpeg_libpng-bin-win32-vs81.zip.

Cheers,
Phil.

Re: [openvrml-develop] OpenVRML 0.16.6 on Windows.. still no success.

[Braden M. <br...@en...>]

On Tue, 2007-10-23 at 15:54 +1300, Philip Lamb wrote:
> Hi all,
> 
> If anyone here is successfully using OpenVRML-0.16.6 on Windows (e.g.  
> can open a .wrl file with sdl-viewer.exe), I'd like to hear how you  
> did it. Getting it working has been a labyrinthine mess. I can  
> compile OK (if OK means with 500+ warnings) but actually viewing  
> anything is an unreached goal.

I have done it.

In general, the warnings are there because they are silly and because
disabling them doesn't really work. If there are warnings you believe to
be pointing to legitimate issues, it is possible you have found
something that I overlooked in the sea of worthless spew (and something
that gcc does not complain about). In that case, please file a bug in
SourceForge.

> Here are the steps so far:
> 1) try to compile boost on Windows. Give up. Use boost- 
> consulting.com's somewhat broken boost binaries installer. OK.
> 2) try to compile openvrml against gnuwin32's freetype/zlib/libpng/ 
> jpeg. C-runtime model incompatibility forces download and build zlib,  
> jpeg, libpng, freetype from source. Oh joy.

It's been a while since I set up libpng and libjpeg to build with on
Windows. I'm guilty of punting here. Can you tell me more about the C
runtime incompatibilities?

I'll Feel The Pain and get this set up.

I have built with FreeType fairly recently. It's not hard to build from
source on Windows.

> 3) try to compile openvrml. Fix issues with openvrml.vsproj file,  
> including include paths picking up linux openvrml-config instead of  
> windows openvrml config.

That's not a problem with the default include paths. It's a problem with
putting platform-specific build files in a source tree that you're
trying to use with multiple platforms. In other words, it's a problem
with building in the srcdir. Don't Do That.

> 4) compile openvrml, openvrml-gl, sdl-viewer.
> 5) TEST code (ARToolKit's libARvrml and simpleVRML example, which  
> currently works flawlessly with openvrml-0.16.6 on Linux and Mac OS  
> X). Bails out with "unrecognized error during resolution" "no  
> alternative URI could be resolved".
> 6) TEST code using sdl-viewer.exe and relative path to .wrl file.  
> Bails out with "invalid URI". Same result with absolute path.  
> Put .wrl file in same folder as sdl-viewer; crashes.

Can you please file a bug in SourceForge and attach the .wrl file you're
using? I will take a look at this.

-- 
Braden McDaniel                           e-mail: <br...@en...>
<http://endoframe.com>                    Jabber: <br...@ja...>

[openvrml-develop] OpenVRML 0.16.6 on Windows.. still no success.

[Philip L. <ph...@ed...>]

Hi all,

If anyone here is successfully using OpenVRML-0.16.6 on Windows (e.g.  
can open a .wrl file with sdl-viewer.exe), I'd like to hear how you  
did it. Getting it working has been a labyrinthine mess. I can  
compile OK (if OK means with 500+ warnings) but actually viewing  
anything is an unreached goal.

Here are the steps so far:
1) try to compile boost on Windows. Give up. Use boost- 
consulting.com's somewhat broken boost binaries installer. OK.
2) try to compile openvrml against gnuwin32's freetype/zlib/libpng/ 
jpeg. C-runtime model incompatibility forces download and build zlib,  
jpeg, libpng, freetype from source. Oh joy.
3) try to compile openvrml. Fix issues with openvrml.vsproj file,  
including include paths picking up linux openvrml-config instead of  
windows openvrml config.
4) compile openvrml, openvrml-gl, sdl-viewer.
5) TEST code (ARToolKit's libARvrml and simpleVRML example, which  
currently works flawlessly with openvrml-0.16.6 on Linux and Mac OS  
X). Bails out with "unrecognized error during resolution" "no  
alternative URI could be resolved".
6) TEST code using sdl-viewer.exe and relative path to .wrl file.  
Bails out with "invalid URI". Same result with absolute path.  
Put .wrl file in same folder as sdl-viewer; crashes.

At this stage, I am ready to give up on making libARvrml available on  
Windows... unless perhaps someone with more time on their hands than  
me has persisted with this and has it working.

Regards,
Phil.

[openvrml-develop] Spirit parser work in CVS

[Braden M. <br...@en...>]

I have started committing my changes toward replacing OpenVRML's
Antlr-based parsers with Spirit-based ones on the
OpenVRML-Spirit-BRANCH.

As the code currently stands, the new VRML97 parser is in place and
working. (At least to the extent that it can load rotation_toy.wrl
successfully.) I am currently working on the X3D-VRML one.

-- 
Braden McDaniel                           e-mail: <br...@en...>
<http://endoframe.com>                    Jabber: <br...@ja...>

Re: [openvrml-develop] Building OpenVRML on OpenSuSE 10.2

[Braden M. <br...@en...>]

On Thu, 2007-10-11 at 17:00 +0200, Gino van den Bergen wrote:
> I'm trying to build OpenVRML 0.16.6 under OpenSuSE 10.2. My problem is 
> that the configure script fails with the following message:
> 
> configure: error: SpiderMonkey is required for Script node JavaScript 
> support.
> 
> Obviously I need to install the Mozilla JS engine, however I do not seem 
> to be able to find it in the otherwise very complete OpenSuSE 10.2 
> distribution.

Googling a bit, mozilla-xulrunner181-devel looks like it's probably what
you need.

Unfortunately, OpenVRML 0.16.6's configure doesn't know about the
xulrunner* variants of the pkg-config files. And while you could
override JS_CFLAGS and JS_LIBS, configure still tries to use pkg-config
with the metadata file in order to establish the directory that should
be used in the rpath flag. Unfortunately I didn't have the foresight to
make GRE_PATH an AC_ARG_VAR. :-/

Would you be able to test a patch?

>  I've tried to build SpiderMonkey from source however the 
> makefile fails on the install.  Rather than waste another day on trying 
> to fix this I would like  to ask you guys whether it is possible to 
> create a config flag for disabling the use of JavaScript an simply  fix 
> the configure script.

That flag already exists: --disable-script-node-javascript.

And, BTW, if you were able to build the stand-alone SpiderMonkey
distribution --with-libjs will make OpenVRML's configure look for it
instead of the libmozjs version that gets included with Firefox,
XULrunner, etc.

>  I need OpenVRML for loading  wrl files and I do 
> not need to build a full-fledged VRML browser, so I probably can live 
> without  script nodes. Alternatively, perhaps the OpenVRML team could 
> include the latest SpiderMonkey source in the OpenVRML  distribution. 
> This would certainly  save a lot of people from the hassle  of mucking 
> about with getting the proper  libs at the proper place.

Not gonna happen. Tracking 3rd-party sources is no fun, so I do it as
little as possible.

> Also, I would like to use my own renderer and need to translate the 
> internal representation from OpenVRML to my own. Are there any examples 
> of how this can be done?

There are a couple of ways to skin the cat. The most straightforward way
would be to implement your own viewer. In that case, the existing OpenGL
renderer (openvrml::gl::viewer) should be a pretty good example.

The other way would be to use openvrml::node_traverser to get the data
from nodes and do whatever you want with it. You should be able to find
some examples of node_traverser uses in the libopenvrml source.

The viewer approach should be faster, but less flexible.

-- 
Braden McDaniel                           e-mail: <br...@en...>
<http://endoframe.com>                    Jabber: <br...@ja...>

[openvrml-develop] Building OpenVRML on OpenSuSE 10.2

[Gino v. d. B. <gva...@vi...>]

I'm trying to build OpenVRML 0.16.6 under OpenSuSE 10.2. My problem is 
that the configure script fails with the following message:

configure: error: SpiderMonkey is required for Script node JavaScript 
support.

Obviously I need to install the Mozilla JS engine, however I do not seem 
to be able to find it in the otherwise very complete OpenSuSE 10.2 
distribution. I've tried to build SpiderMonkey from source however the 
makefile fails on the install.  Rather than waste another day on trying 
to fix this I would like  to ask you guys whether it is possible to 
create a config flag for disabling the use of JavaScript an simply  fix 
the configure script. I need OpenVRML for loading  wrl files and I do 
not need to build a full-fledged VRML browser, so I probably can live 
without  script nodes. Alternatively, perhaps the OpenVRML team could 
include the latest SpiderMonkey source in the OpenVRML  distribution. 
This would certainly  save a lot of people from the hassle  of mucking 
about with getting the proper  libs at the proper place.

Also, I would like to use my own renderer and need to translate the 
internal representation from OpenVRML to my own. Are there any examples 
of how this can be done?

Thanks,

Gino

Re: [openvrml-develop] build failure under -release on windows

[Philip L. <ph...@ed...>]

On 17/09/2007, at 2:10 PM, Braden McDaniel wrote:
>
> That looks like what you get if a configure-generated openvrml- 
> config.h
> is in your include path ahead of the one in
> ide-projects/Windows/VisualC8_0/OpenVRML/openvrml.
>

Well spotted.. after trying unsucessfully to get it to build by  
making the release build settings match the debug ones, your tip lead  
me to look at the include path. The include paths are in a different  
order on the release build. Fixed that, and now it builds. Next  
thing.. testing.

Cheers,

Phil

Re: [openvrml-develop] build failure in vrml97node.cpp, 0.16.6 on Mac OS X

[Braden M. <br...@en...>]

On Mon, 2007-09-17 at 15:59 +1200, Philip Lamb wrote:
> Hi all,
> 
> Here is another obstacle.. hopefully an easy fix for this one..
> 
> Building OpenVRML-0.16.6, this time on Mac OS X 10.4.10 (Xcode tools  
> 2.4.1, gcc v4.0.1.Dependencies provided by Fink; gnome etc.,  
> seamonkey (for libjs), libpng/jpeg/z.
> 
> The error appears to be a "const" qualifier popping up unexpectedly.  

Or rather, one not appearing where it is expected.

> Here is the relevant part of the build log:

[snip]

OPENVRML_FT_CONST needs to get defined to "const" (for recent versions
of FreeType). configure should have figured this out; but apparently the
test didn't do what it was supposed to.

I have a hunch as to why this is. If I'm right, the attached patch to
configure.ac should do the trick. (Run autoreconf to regenerate
configure after applying it. If you don't have autoconf available, let
me know and I will create a patch to configure for you.)

If it still doesn't work, please send your config.log.

-- 
Braden McDaniel                           e-mail: <br...@en...>
<http://endoframe.com>                    Jabber: <br...@ja...>

[openvrml-develop] build failure in vrml97node.cpp, 0.16.6 on Mac OS X

[Philip L. <ph...@ed...>]

Hi all,

Here is another obstacle.. hopefully an easy fix for this one..

Building OpenVRML-0.16.6, this time on Mac OS X 10.4.10 (Xcode tools  
2.4.1, gcc v4.0.1.Dependencies provided by Fink; gnome etc.,  
seamonkey (for libjs), libpng/jpeg/z.

The error appears to be a "const" qualifier popping up unexpectedly.  
Here is the relevant part of the build log:

/bin/sh ../../libtool --tag=CXX   --mode=compile g++ -DHAVE_CONFIG_H - 
I. -I../..  -I../../lib/antlr -I../../java -I../../src/libopenvrml - 
DOPENVRML_LIBDIR_=\"/usr/local/lib\" -DOPENVRML_PKGDATADIR_=\"/usr/ 
local/share/openvrml\" -DBOOST_SPIRIT_THREADSAFE - 
DBOOST_MPL_CFG_NO_PREPROCESSED_HEADERS - 
DBOOST_MPL_LIMIT_VECTOR_SIZE=30 -I/sw/include -D_THREAD_SAFE -I/usr/ 
X11R6/include -I/sw/lib/freetype219/include/freetype2 -I/sw/lib/ 
freetype219/include -I/sw/include/js -DXP_UNIX -g -O2 -c -o openvrml/ 
vrml97node.lo openvrml/vrml97node.cpp
  g++ -DHAVE_CONFIG_H -I. -I../.. -I../../lib/antlr -I../../java - 
I../../src/libopenvrml -DOPENVRML_LIBDIR_=\"/usr/local/lib\" - 
DOPENVRML_PKGDATADIR_=\"/usr/local/share/openvrml\" - 
DBOOST_SPIRIT_THREADSAFE -DBOOST_MPL_CFG_NO_PREPROCESSED_HEADERS - 
DBOOST_MPL_LIMIT_VECTOR_SIZE=30 -I/sw/include -D_THREAD_SAFE -I/usr/ 
X11R6/include -I/sw/lib/freetype219/include/freetype2 -I/sw/lib/ 
freetype219/include -I/sw/include/js -DXP_UNIX -g -O2 -c openvrml/ 
vrml97node.cpp  -fno-common -DPIC -o openvrml/.libs/vrml97node.o
In file included from /sw/include/jpeglib.h:28,
                  from openvrml/vrml97node.cpp:34:
/sw/include/jconfig.h:12:1: warning: "HAVE_STDLIB_H" redefined
In file included from openvrml/vrml97node.cpp:26:
../../config.h:23:1: warning: this is the location of the previous  
definition
openvrml/vrml97node.cpp: In member function  
'void<unnamed>::text_node::update_geometry()':
openvrml/vrml97node.cpp:22976: error: invalid conversion from 'int (*) 
(FT_Vector*, void*)' to 'int (*)(const FT_Vector*, void*)'
openvrml/vrml97node.cpp:22976: error: invalid conversion from 'int (*) 
(FT_Vector*, void*)' to 'int (*)(const FT_Vector*, void*)'
openvrml/vrml97node.cpp:22976: error: invalid conversion from 'int (*) 
(FT_Vector*, FT_Vector*, void*)' to 'int (*)(const FT_Vector*, const  
FT_Vector*, void*)'
openvrml/vrml97node.cpp:22976: error: invalid conversion from 'int (*) 
(FT_Vector*, FT_Vector*, FT_Vector*, void*)' to 'int (*)(const  
FT_Vector*, const FT_Vector*, const FT_Vector*, void*)'
make[4]: *** [openvrml/vrml97node.lo] Error 1
make[3]: *** [all] Error 2
make[2]: *** [all-recursive] Error 1
make[1]: *** [all-recursive] Error 1
make: *** [all] Error 2


My configure line was:
./configure --with-apple-opengl-framework --disable-dependency- 
tracking FREETYPE_CFLAGS="`/sw/lib/freetype219/bin/freetype-config -- 
cflags`" FREETYPE_LIBS="`/sw/lib/freetype219/bin/freetype-config -- 
libs`" JS_CFLAGS="-I/sw/include/js -DXP_UNIX" CPPFLAGS="-I/sw/ 
include" LDFLAGS="-L/sw/lib" --disable-mozilla-plugin

Cheers,
Phil.

Re: [openvrml-develop] build failure under -release on windows

[Braden M. <br...@en...>]

On Mon, 2007-09-17 at 11:27 +1200, Philip Lamb wrote:
> Hi all,
> 
> I'm doing a build of openvrml-0.16.6 on Windows XP SP 2, under MS VS  
> 2005 (first release, not VS2005 SP 1) and the build of openvrml-gl is  
> failing under the -release config. (-debug config builds OK).
> 
> Here is the error log.. not a lot to go on here, but might mean  
> something to c++ / MSVS gurus (I am neither).

[snip]

That looks like what you get if a configure-generated openvrml-config.h
is in your include path ahead of the one in
ide-projects/Windows/VisualC8_0/OpenVRML/openvrml.

-- 
Braden McDaniel                           e-mail: <br...@en...>
<http://endoframe.com>                    Jabber: <br...@ja...>

[openvrml-develop] build failure under -release on windows

[Philip L. <ph...@ed...>]

Hi all,

I'm doing a build of openvrml-0.16.6 on Windows XP SP 2, under MS VS  
2005 (first release, not VS2005 SP 1) and the build of openvrml-gl is  
failing under the -release config. (-debug config builds OK).

Here is the error log.. not a lot to go on here, but might mean  
something to c++ / MSVS gurus (I am neither).

------ Build started: Project: openvrml-gl, Configuration: Release  
Win32 ------
Compiling...
viewer.cpp
c:\Projects\openvrml-0.16.6\src\libopenvrml\openvrml/basetypes.h 
(45) : error C3861: 'visibility': identifier not found
c:\Projects\openvrml-0.16.6\src\libopenvrml\openvrml/basetypes.h 
(45) : error C4430: missing type specifier - int assumed. Note: C++  
does not support default-int
c:\Projects\openvrml-0.16.6\src\libopenvrml\openvrml/basetypes.h 
(45) : error C2143: syntax error : missing ';' before 'std::istream'
c:\Projects\openvrml-0.16.6\src\libopenvrml\openvrml/basetypes.h 
(45) : error C2143: syntax error : missing ';' before '&'
c:\Projects\openvrml-0.16.6\src\libopenvrml\openvrml/basetypes.h 
(45) : error C4430: missing type specifier - int assumed. Note: C++  
does not support default-int
c:\Projects\openvrml-0.16.6\src\libopenvrml\openvrml/basetypes.h 
(45) : error C2377: 'std::istream' : redefinition; typedef cannot be  
overloaded with any other symbol
         C:\Program Files\Microsoft Visual Studio 8\VC\include\iosfwd 
(700) : see declaration of 'std::istream'
c:\Projects\openvrml-0.16.6\src\libopenvrml\openvrml/basetypes.h 
(45) : error C2888: 'std::basic_istream<_Elem,_Traits>  
std::istream' : symbol cannot be defined within namespace 'openvrml'
         with
         [
             _Elem=char,
             _Traits=std::char_traits<char>
         ]
c:\Projects\openvrml-0.16.6\src\libopenvrml\openvrml/basetypes.h 
(45) : error C2065: 'in' : undeclared identifier
c:\Projects\openvrml-0.16.6\src\libopenvrml\openvrml/basetypes.h 
(45) : error C2065: 'c' : undeclared identifier
c:\Projects\openvrml-0.16.6\src\libopenvrml\openvrml/basetypes.h 
(45) : error C2275: 'openvrml::color' : illegal use of this type as  
an expression
         c:\Projects\openvrml-0.16.6\src\libopenvrml\openvrml/ 
basetypes.h(43) : see declaration of 'openvrml::color'
c:\Projects\openvrml-0.16.6\src\libopenvrml\openvrml/basetypes.h 
(45) : error C4430: missing type specifier - int assumed. Note: C++  
does not support default-int
c:\Projects\openvrml-0.16.6\src\libopenvrml\openvrml/basetypes.h 
(45) : fatal error C1903: unable to recover from previous error(s);  
stopping compilation
Build log was saved at "file://c:\Projects\openvrml-0.16.6\ide- 
projects\Windows\VisualC8_0\OpenVRML\openvrml-gl\Release\BuildLog.htm"
openvrml-gl - 12 error(s), 0 warning(s)

Can anyone offer any hints?

Cheers,

Phil

Re: [openvrml-develop] Toward 0.17.0: replacing ANTLR with Spirit

[Braden M. <br...@en...>]

On Wed, 2007-09-05 at 08:55 -0400, Reed Hedges wrote:
> Sounds good. I'd be interested to hear about your experience with Spirit. I've
> read through the docs a bit but haven't tried it yet, though am thinking of
> using it for some stuff.

Once you become acquainted with how Spirit wants you to do things, it
becomes apparent that there is a whole lot of flexibility built into the
model. Arbitrary semantic checks are really easy to do. However, in
general, "easy" doesn't mean "short to type". Spirit is verbose. The
main reason for this is that C++'s syntax for function objects is
verbose, and Spirit relies heavily on function objects.

August got really busy and I wasn't able to spend much time on this
stuff. I also had to take some time out to fix some openvrml-player
issues. I'm back at it, but I'm still in the process of integrating the
new VRML97 parser.

-- 
Braden McDaniel                           e-mail: <br...@en...>
<http://endoframe.com>                    Jabber: <br...@ja...>

Re: [openvrml-develop] Toward 0.17.0: replacing ANTLR with Spirit

[Reed H. <re...@in...>]

Sounds good. I'd be interested to hear about your experience with Spirit. I've
read through the docs a bit but haven't tried it yet, though am thinking of
using it for some stuff.

Reed


On Thu, Aug 02, 2007 at 01:43:30AM -0400, Braden McDaniel wrote:
> I've decided to replace the ANTLR-based parsers in OpenVRML with
> Spirit-based ones for the 0.17.0 release.

[openvrml-develop] Toward 0.17.0: replacing ANTLR with Spirit

[Braden M. <br...@en...>]

I've decided to replace the ANTLR-based parsers in OpenVRML with
Spirit-based ones for the 0.17.0 release.

There is plenty of motivation for this:

      * Using Spirit means there is no code generation step.
      * Spirit is included with Boost; OpenVRML already requires Boost,
        so using Spirit doesn't really add a dependency.
      * Spirit is entirely a template library; so there is no run-time
        library to link with.
      * C++ has been a second-class citizen for ANTLR; and there's no
        sign of that changing. If anything, things are a bit worse in
        this regard with ANTLR 3.0.
      * The release of ANTLR 3.0 means it's time either to upgrade or to
        migrate away from ANTLR. I've chosen the latter.

Now, all that said, I have made significant progress in this direction.
As of this writing, I have a Spirit-based parser for VRML97 that is at
least as strict as the existing ANTLR parser.

And perhaps the best thing about it is that its semantic actions are
pluggable. That means it makes sense to expose the parser itself to
users; users can invoke the parser with arbitrary semantic actions. This
should be extremely useful to persons who just want to read in some VRML
and do something arbitrary with it. Even better, the parser is entirely
in headers--users who use no other part of OpenVRML will not need to
link with libopenvrml.

I haven't committed anything to CVS just yet. I'd like to finish
migrating libopenvrml to use the new parser before doing that. But that
shouldn't be too far away at this point.

-- 
Braden McDaniel                           e-mail: <br...@en...>
<http://endoframe.com>                    Jabber: <br...@ja...>

Re: [openvrml-develop] Migrating to Subversion

[Braden M. <br...@en...>]

On Tue, 2007-05-15 at 18:10 -0400, Braden McDaniel wrote:

[snip]

> With this in mind, I'm inclined to migrate OpenVRML to Subversion in the 
> near future; probably shortly after 0.16.5 is released.

Just to follow up on this, I've reconsidered a bit and my current plan
is to do this migration immediately after releasing 0.17.0. (Which,
barring some bad bug in 0.16.6 warranting a 0.16.7, should be the next
OpenVRML release.)

-- 
Braden McDaniel                           e-mail: <br...@en...>
<http://endoframe.com>                    Jabber: <br...@ja...>

[openvrml-develop] ANN: OpenVRML 0.16.6

[Braden M. <br...@en...>]

OpenVRML 0.16.6 is now available.  The distribution can be obtained
from

  <http://downloads.sourceforge.net/openvrml/openvrml-0.16.6.tar.gz>

OpenVRML is a C++ runtime library for VRML97 and X3D worlds.  It is
capable of reading and displaying VRML/X3D; it can be used for creating
loaders, file converters, and VRML/X3D browsers.

OpenVRML includes an out-of-process viewer component for use in X11
environments along with hosts for this component in the form of a
Mozilla browser plug-in and a stand-alone player.

You can find OpenVRML on the Web at

  <http://openvrml.org>


New in OpenVRML 0.16.6:

- Fixed JavaScript support to work in the presence of JS_THREADSAFE.

- Added a configure option --with-libjs to accommodate compiling
  against a stand-alone SpiderMonkey distribution.

Re: [openvrml-develop] How to build OpenVRML on VisualStudio 2005 and 2003

[Braden M. <br...@en...>]

Larry Mullen wrote:
> Braden,
> 
> Thanks for all the help. I've also been able to build it with Visual
> Studio .Net 2003.

You're welcome.

> On 7/10/07, Braden McDaniel <br...@en...> wrote:
>> On Mon, 2007-07-09 at 10:51 -0400, Braden McDaniel wrote:
>>> On Mon, 2007-07-09 at 09:37 -0400, Larry Mullen wrote:
>>>> Thanks again Braden. I was able to build openvrml and everything else
>>>> in the project. I really appreciate your help.
>>>>
>>>> Couple more questions
>>>>
>>>> 1. If I run sdl_viewer.cpp with a valid url (example
>>>> http://www.3dtrue.com/vrml/wrl/vrmlelevator.wrl) I get the following
>>>> error
>>>>
>>>> unreachable URI
>>>> no alternative URI could be resolved
>>> Hm; could be a bug. I'll look into this.
>> I'm an idiot. Of course that doesn't work. sdl-viewer only supports file
>> URLs.

As it turns out, there is a minor bug here. You should have seen a 
different, more informative error message. (Note line 203 of 
sdl_viewer.cpp; unfortunately this exception is getting swallowed 
without emitting that error message.)

I will fix this for 0.16.6.

Braden

Re: [openvrml-develop] How to build OpenVRML on VisualStudio 2005 and 2003

[Larry M. <lar...@gm...>]

Braden,

Thanks for all the help. I've also been able to build it with Visual
Studio .Net 2003.



On 7/10/07, Braden McDaniel <br...@en...> wrote:
> On Mon, 2007-07-09 at 10:51 -0400, Braden McDaniel wrote:
> > On Mon, 2007-07-09 at 09:37 -0400, Larry Mullen wrote:
> > > Thanks again Braden. I was able to build openvrml and everything else
> > > in the project. I really appreciate your help.
> > >
> > > Couple more questions
> > >
> > > 1. If I run sdl_viewer.cpp with a valid url (example
> > > http://www.3dtrue.com/vrml/wrl/vrmlelevator.wrl) I get the following
> > > error
> > >
> > > unreachable URI
> > > no alternative URI could be resolved
> >
> > Hm; could be a bug. I'll look into this.
>
> I'm an idiot. Of course that doesn't work. sdl-viewer only supports file
> URLs.
>
> --
> Braden McDaniel                           e-mail: <br...@en...>
> <http://endoframe.com>                    Jabber: <br...@ja...>
>
>
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by DB2 Express
> Download DB2 Express C - the FREE version of DB2 express and take
> control of your XML. No limits. Just data. Click to get it now.
> http://sourceforge.net/powerbar/db2/
> _______________________________________________
> openvrml-develop mailing list
> ope...@li...
> https://lists.sourceforge.net/lists/listinfo/openvrml-develop
>

Re: [openvrml-develop] How to build OpenVRML on VisualStudio 2005 and 2003

[Braden M. <br...@en...>]

On Mon, 2007-07-09 at 10:51 -0400, Braden McDaniel wrote:
> On Mon, 2007-07-09 at 09:37 -0400, Larry Mullen wrote:
> > Thanks again Braden. I was able to build openvrml and everything else
> > in the project. I really appreciate your help.
> > 
> > Couple more questions
> > 
> > 1. If I run sdl_viewer.cpp with a valid url (example
> > http://www.3dtrue.com/vrml/wrl/vrmlelevator.wrl) I get the following
> > error
> > 
> > unreachable URI
> > no alternative URI could be resolved
> 
> Hm; could be a bug. I'll look into this.

I'm an idiot. Of course that doesn't work. sdl-viewer only supports file
URLs.

-- 
Braden McDaniel                           e-mail: <br...@en...>
<http://endoframe.com>                    Jabber: <br...@ja...>

Re: [openvrml-develop] How to build OpenVRML on VisualStudio 2005 and 2003

[Braden M. <br...@en...>]

On Mon, 2007-07-09 at 09:37 -0400, Larry Mullen wrote:
> Thanks again Braden. I was able to build openvrml and everything else
> in the project. I really appreciate your help.
> 
> Couple more questions
> 
> 1. If I run sdl_viewer.cpp with a valid url (example
> http://www.3dtrue.com/vrml/wrl/vrmlelevator.wrl) I get the following
> error
> 
> unreachable URI
> no alternative URI could be resolved

Hm; could be a bug. I'll look into this.

> 2. If I give the url as "file:///C:\temp\001\001.wrl" (which exists) I
> get the following error
> 
> invalid URI
> no alternative URI could be resolved

This is working as designed; URIs don't have '\' in them. Try using '/'
instead.

> 3. As you mentioned as reply to my initial questions, to verify that a
> VRML is constructed correctly, I need to load it into a scene. So if
> the following code does not throw an exception, that implies my VRML
> is contructed properly, correct?
> 
> try {
>         using std::string;
>         using std::vector;
>         const string url = argv[1];
> 
>         sdl_viewer v(url);
>         browser b;
> 
>         vector<string> uri(1, url);
>         vector<string> parameter;
>         b.load_url(uri, parameter);
> 	  openvrml::scene s(b);
> 		
> 	  s.load(*s.get_resource(uri).get());
> 		
> 		
> 
> 
>     } catch (std::exception & ex) {
>         cerr << ex.what() << endl;
>         return EXIT_FAILURE;
>     }

That's not very efficient; you end up loading the file twice.
browser::load_url should be all you need. The problem, though, is that
browser::load_url does not block. If you use it and you want to confirm
it succeeded, you can add a browser_listener for the initialized event.

However, it's probably easier for you just to use browser::set_world
instead of browser::load_url. set_world blocks; and if it finishes
without throwing, you're good.

-- 
Braden McDaniel                           e-mail: <br...@en...>
<http://endoframe.com>                    Jabber: <br...@ja...>

Re: [openvrml-develop] How to build OpenVRML on VisualStudio 2005 and 2003

[Larry M. <lar...@gm...>]

Thanks again Braden. I was able to build openvrml and everything else
in the project. I really appreciate your help.

Couple more questions

1. If I run sdl_viewer.cpp with a valid url (example
http://www.3dtrue.com/vrml/wrl/vrmlelevator.wrl) I get the following
error

unreachable URI
no alternative URI could be resolved

2. If I give the url as "file:///C:\temp\001\001.wrl" (which exists) I
get the following error

invalid URI
no alternative URI could be resolved

3. As you mentioned as reply to my initial questions, to verify that a
VRML is constructed correctly, I need to load it into a scene. So if
the following code does not throw an exception, that implies my VRML
is contructed properly, correct?

try {
        using std::string;
        using std::vector;
        const string url = argv[1];

        sdl_viewer v(url);
        browser b;

        vector<string> uri(1, url);
        vector<string> parameter;
        b.load_url(uri, parameter);
	  openvrml::scene s(b);
		
	  s.load(*s.get_resource(uri).get());
		
		


    } catch (std::exception & ex) {
        cerr << ex.what() << endl;
        return EXIT_FAILURE;
    }

Thanks again for all your help,

Rohit.

On 7/3/07, Braden McDaniel <br...@en...> wrote:
> Larry Mullen wrote:
> > Thanks again Braden - it's too late to chicken out - so I built Boost
> > anyway....
> >
> > now I get the following linker errors....
>
> [snip]
>
> Those are FreeType symbols. Check to make sure the FreeType lib you are
> linking against actually exists. FreeType makes new releases frequently
> and the library names are versioned; so the library built by the version
> of FreeType you downloaded might have a different name than the one
> referenced in the project file that comes with OpenVRML 0.16.5.
>
> Also note that FreeType is only necessary for *rendering* text. If you
> are only interested in reading in a file to verify correctness, you
> probably don't care about that. If you don't define
> OPENVRML_ENABLE_RENDER_TEXT_NODE, you don't need FreeType. See README
> ("Significant preprocessor symbols") for more information.
>
> Braden
>
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by DB2 Express
> Download DB2 Express C - the FREE version of DB2 express and take
> control of your XML. No limits. Just data. Click to get it now.
> http://sourceforge.net/powerbar/db2/
> _______________________________________________
> openvrml-develop mailing list
> ope...@li...
> https://lists.sourceforge.net/lists/listinfo/openvrml-develop
>

[openvrml-develop] 0.16.6 coming soon

[Braden M. <br...@en...>]

Over the weekend I discovered that, once again, the Mozilla folks have
changed the flags required for building against their stuff without
updating the flags in their pkg-config metadata files.

To be fair, that's just one aspect of the problem. It turns out OpenVRML
also needs some code-level tweaks to work with the version of JavaScript
that is included with Firefox 2.0.

Briefly: JavaScript run in OpenVRML built against Firefox 2.0's libmozjs
is extremely likely (well, probably certain) to result in a crash.

For those interested in the details: libmozjs is built with
JS_THREADSAFE. I don't know how long that's been true; I would guess
it's new to the version of SpiderMonkey included with Firefox 2; but
that might not be the case. What is definitely the case is that this
version of SpiderMonkey--when compiled with JS_THREADSAFE--requires that
client code be compiled with JS_THREADSAFE as well. Furthermore, client
code that is compiled as such must also apply the functions
JS_SetContextThread and JS_ClearContextThread appropriately.

This was great fun to chase down starting from clue 0. :-/

But I seem to have this working locally; and I'll check in the fix soon.
And unfortunately, this warrants a 0.16.6 release.

-- 
Braden McDaniel                           e-mail: <br...@en...>
<http://endoframe.com>                    Jabber: <br...@ja...>

Re: [openvrml-develop] How to build OpenVRML on VisualStudio 2005 and 2003

[Braden M. <br...@en...>]

Larry Mullen wrote:
> Thanks again Braden - it's too late to chicken out - so I built Boost 
> anyway....
> 
> now I get the following linker errors....

[snip]

Those are FreeType symbols. Check to make sure the FreeType lib you are 
linking against actually exists. FreeType makes new releases frequently 
and the library names are versioned; so the library built by the version 
of FreeType you downloaded might have a different name than the one 
referenced in the project file that comes with OpenVRML 0.16.5.

Also note that FreeType is only necessary for *rendering* text. If you 
are only interested in reading in a file to verify correctness, you 
probably don't care about that. If you don't define 
OPENVRML_ENABLE_RENDER_TEXT_NODE, you don't need FreeType. See README 
("Significant preprocessor symbols") for more information.

Braden