Re: [osg-users] Compilation error with MSVC 2015

[Jan Ciger]

Hello again,

On Mon, Jul 4, 2016 at 4:29 PM, Robert Osfield 
wrote:

> Weird.  I have just had a look t the source file the compiler is
> complaining about and it looks pretty standard.
>
> The gles plugin on Linux is not particularly large either. I can't see
> any obvious reasons why there would be any need for some special
> options.
>
>
This is what the doc says about /bigobj:
https://msdn.microsoft.com/en-us/library/ms173499.aspx

My suspicion is that as that optimizer code works over all possible OSG
nodes, it is possible it is trying to instantiate ton of templates (e.g.
osg::ref_ptr<>), pushing the number of sections over 64k in my particular
case, with my set of self-compiled dependencies.

It seems that adding /bigobj should be safe, only compilers older than MSVC
2005 wouldn't be able to deal with such object files - which is a
non-issue, because one cannot really mix C++ object files/libs between MSVC
versions safely anyway.

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

Re: [osg-users] Compilation error with MSVC 2015

[Jan Ciger]

Hello,

On Mon, Jul 4, 2016 at 4:29 PM, Robert Osfield 
wrote:

> Hi Jan,
>
> On 4 July 2016 at 14:12, Jan Ciger  wrote:
> > It was a fresh checkout from Github master from a few minutes ago. I have
> > modified the main CMakeLists.txt and added /bigobj, as mentioned in the
> > error message, to CMAKE_CXX_FLAGS for MSVC and now the same code seems to
> > build fine,so that doesn't look like a corrupted file. I was building a
> > debug build when I have got that error, not sure whether it has any
> > relevance.
>
> Weird.  I have just had a look t the source file the compiler is
> complaining about and it looks pretty standard.
>
> The gles plugin on Linux is not particularly large either. I can't see
> any obvious reasons why there would be any need for some special
> options.
>
> What size are the object files and resulting osgdb_gles.dll?
>


The OpenGLESGeometryOptimizer.obj is 4.84MB in Release and 28.81MB in
Debug, the remaining files are few hundred kB (Release), few megs (in
Debug).

The DLL is 770kB in release and 3.84MB in debug.

I have noticed that the incremental linking was off in the CMakeLists.txt
for MSVC, that could perhaps have an impact.

Can the option just been added to the gles pugin if absolutely required?
>

Yes, possibly. However, it may be that the build would break elsewhere too
- I didn't have time to rebuild the whole mess several times today, so I
have put it only globally. I don't think that the GLES code is somehow
special.


>
> Right now, it seems pretty odd that the gles plugin would break some
> compiler limits that we've never hit before in any other module or
> plugins, so I'm inclined to think there is something specific going
> on, perhaps a bug with the compiler that is throwing things off.
>

Yes, I wouldn't be surprised. Or it could be triggering generation of ton
of code in 2015 that exceeds the limits. MSVC 2015 broke a lot of code that
was compiling before, because it is both stricter and updated for newer C++
standard. So when rebuilding older C++ code I am now often troubleshooting
compilation failures about code referencing deleted functions, template
problems, you name it.

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

Re: [osg-users] Compilation error with MSVC 2015

[Robert Osfield]

Hi Jan,

On 4 July 2016 at 14:12, Jan Ciger  wrote:
> It was a fresh checkout from Github master from a few minutes ago. I have
> modified the main CMakeLists.txt and added /bigobj, as mentioned in the
> error message, to CMAKE_CXX_FLAGS for MSVC and now the same code seems to
> build fine,so that doesn't look like a corrupted file. I was building a
> debug build when I have got that error, not sure whether it has any
> relevance.

Weird.  I have just had a look t the source file the compiler is
complaining about and it looks pretty standard.

The gles plugin on Linux is not particularly large either. I can't see
any obvious reasons why there would be any need for some special
options.

What size are the object files and resulting osgdb_gles.dll?

> I have yet to see whether the extra flag will have any runtime consequences.

Can the option just been added to the gles pugin if absolutely required?

Right now, it seems pretty odd that the gles plugin would break some
compiler limits that we've never hit before in any other module or
plugins, so I'm inclined to think there is something specific going
on, perhaps a bug with the compiler that is throwing things off.

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

Re: [osg-users] Compilation error with MSVC 2015

[Jan Ciger]

Hello,

On Mon, Jul 4, 2016 at 2:29 PM, Robert Osfield 
wrote:

> Hi Jan,
>
> The gles plugin is a tool to aid conversion of fixed function pipeline
> scene graphs across to shader based ones compatible with GLES2, it's
> not specifically tied to GLES though and should happily compile and
> work with a normal OpenGL build.
>

OK.


>
> The error you report looks like a compiler bug/error rather than a
> code error.  The only other thing I can think of that for some reason
> your OpenGLESGeometryOptimizer.cpp file has got corrupted.
>
> You don't specify which version of the OSG you are using so can't
> check to see if the file is actually OK.
>


It was a fresh checkout from Github master from a few minutes ago. I have
modified the main CMakeLists.txt and added /bigobj, as mentioned in the
error message, to CMAKE_CXX_FLAGS for MSVC and now the same code seems to
build fine,so that doesn't look like a corrupted file. I was building a
debug build when I have got that error, not sure whether it has any
relevance.

I have yet to see whether the extra flag will have any runtime consequences.

Regards,

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

Re: [osg-users] Compilation error with MSVC 2015

[Robert Osfield]

Hi Jan,

The gles plugin is a tool to aid conversion of fixed function pipeline
scene graphs across to shader based ones compatible with GLES2, it's
not specifically tied to GLES though and should happily compile and
work with a normal OpenGL build.

The error you report looks like a compiler bug/error rather than a
code error.  The only other thing I can think of that for some reason
your OpenGLESGeometryOptimizer.cpp file has got corrupted.

You don't specify which version of the OSG you are using so can't
check to see if the file is actually OK.

Cheers,
Robert.

On 4 July 2016 at 13:23, Jan Ciger  wrote:
> Hello,
>
> I am updating my OSG at work and when building with the MSVC 2015, I am
> getting this:
>
> 72>C:\R\Dependencies-x64-2015\OpenSceneGraph\src\osgPlugins\gles\OpenGLESGeometryOptimizer.cpp
> : fatal error C1128: number of sections exceeded object file format limit:
> compile with /bigobj
>
> And the build fails.
>
> What is strange is that why is it even trying to build that file - I have
> all GLES options off. Has anyone seen this error before?
>
> J.
>
>
> ___
> 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