Hi Daivd,

I have lots to tackle w.r.t submission so my focus will be entirely on
fixes that provide real benefit to users.  I have already reverted the
"warning fix" changes that I feel are hacks, and just from your
explanation of these latest changes know that a number of also of the
type I'd consider bad programming hacks that deserve no place in the
OSG code place.  So to be clear, I believe it's fundamentally wrong to
add code to quieten inappropriate warnings.

If we are having to fix such problems then either the code has a
problem we need to fix and do so in a genuine way - not hacking around
the compiling warning, or to simple disable the problem warnings.

Also please remember every single character change to the OSG has to
potential for breaking it in unexpected ways, be it compile issue or a
runtime issue.  In past widespread warning purges we've introduced
obscure bugs that weren't caught right away.  Only once you see this
issue for real do you perhaps start to realize that one has to tread
carefully and understand that warnings aren't always helpful and can
indeed lead to harmful programming practices.

With this in mind could you review all your changes and asking
yourself can you really justify each of the changes, can you
prioritize the ones that definitely address a real problem that could
cause a bug in the software, or is causing problems with
maintainability. If you feel confident that the change is good then
feel free to submit them.   As I have plenty of other things to do in
the next two days I'll won't look at the rest of your warning fixes,
instead wait for you to vet them and then pass along the ones you feel
are sound given the higher standards that I've laid out above.

Thanks,
Robert.

On 30 June 2013 02:48, David Callu <[email protected]> wrote:
> Hi Robert,
>
> Here the second pass of warning fix
>
> all archive are in the form src/... or include/...  or example/...
>
>
>
> example_refactoring.txz fix warning for osgsidebyside example.
> This example define a class derived from osgGA::GUIEventHandler and
> osg::NodeVisitor.
> This is a very bad example of double inheritance. the warning say that
> osg::Referenced base class
> is inaccessible. I fix this by create 2 class one derived from
> osgGA::GUIEventHandler, a second derived
> from osg::NodeVisitor.
>
>
> osgPlugin_OpenFlight fix some warning and replace ® by (R).
>
>
> other_wrapper_warning.txz fix warning for REGISTER_OBJECT_WRAPPER macro.
> I don't use OSG_UNUSED any more, but I redefined a macro
> REGISTER_EMPTY_OBJECT_WRAPPER
>
>
> wrapper_warning.txz fix warning in osgWrapper when function's parameter is
> not used.
>
>
> to fix warning when variable is used in assert and not in code,
> I define an OSG_ASSERT in osg/Export
>
>
> I now build without warning with GCC and Clang compilator, excepted for
> ffmpeg plugin
>
>
> Cheers
> David
>
> _______________________________________________
> osg-submissions mailing list
> [email protected]
> http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org
>
_______________________________________________
osg-submissions mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org

Reply via email to