Stephan,
I think I've found the issue - it seems that the AvailableReaderWriterIterator
isn't being used safely.
In ReaderWriter::ReadResult Registry::read(const ReadFunctor& readFunctor)
you'll find:
> // first attempt to load the file from existing ReaderWriter's
> AvailableReaderWriterIterator itr(_rwList, _pluginMutex);
> for(;itr.valid();++itr)
> {
> ReaderWriter::ReadResult rr = readFunctor.doRead(*itr);
> if (readFunctor.isValid(rr)) return rr;
> else results.push_back(rr);
> }
and later on…
> // now look for a plug-in to load the file.
> std::string libraryName = createLibraryNameForFile(readFunctor._filename);
> if (loadLibrary(libraryName)!=NOT_LOADED)
> {
> for(;itr.valid();++itr)
> {
> ReaderWriter::ReadResult rr = readFunctor.doRead(*itr);
> if (readFunctor.isValid(rr)) return rr;
> else results.push_back(rr);
> }
> }
The first loop runs through the iterator until itr.valid() returns false.
The second loop never runs because itr.valid() immediately returns false, or
because a compiler optimization assumed that it would return false.
I suspect that when loadLibrary is called, the library is inserted into
_rwList, which would cause itr.valid() to return true, but it appears that the
second call (and the loop it's in) were optimized away.
The following code
> // now look for a plug-in to load the file.
> std::string libraryName = createLibraryNameForFile(readFunctor._filename);
> if (loadLibrary(libraryName)!=NOT_LOADED)
> {
> AvailableReaderWriterIterator itr(_rwList, _pluginMutex);
> for(;itr.valid();++itr)
> {
> ReaderWriter::ReadResult rr = readFunctor.doRead(*itr);
> if (readFunctor.isValid(rr)) return rr;
> else results.push_back(rr);
> }
> }
appears to fix the issue.
There's probably a way to tell the compiler that it can't optimize that call
away (careful placement of a volatile keyword?) but someone more familiar with
that aspect of C++ would have to do it.
This all begs the question: How does one submit a bug report for openscenegraph?
- Ron
On Nov 12, 2013, at 1:01 PM, Ronald Aldrich <[email protected]> wrote:
> Stephan,
>
> I've re-organized my application to match your layout as best I can. It
> appears that I've built a debug build, rather than a release build. I
> haven't been able to figure out how to make a release build using XCode 5
> (This begs the question: What version of XCode and MacOS X are you using?).
>
> One issue I'm running into is that I cannot step and trace anything within
> the OSG library. I'm pretty sure that it's because the modification dates
> between the library as built, and as copied into my application bundle are
> different.
>
> As far as I can tell both the OSG framework and the plugins are loading
> correctly - Still, reading a .OSG file doesn't work.
>
> I think the error messages are a bit of a red herring - I added trail of
> bread crumbs (printf statements) so I could see what's happening.
>
>> FindFileInPath() : USING
>> /Users/raldrich/Documents/fu/osgviewercocoa/DerivedData/osgviewercocoa/Build/Products/Debug/osgviewercocoa.app/Contents/PlugIns/osgdb_osgd.so
>> Opened DynamicLibrary osgPlugins-3.2.0/osgdb_osgd.so
>> LOADED
>> LOADED
>> Warning: Could not find plugin to read objects from file
>> "/Users/raldrich/Documents/fu/OpenSceneGraph/OpenSceneGraph-Data-3.0.0/cessna.osg".
>> 2013-11-12 12:33:43.629 osgviewercocoa[85734:303] File:
>> /Users/raldrich/Documents/fu/OpenSceneGraph/OpenSceneGraph-Data-3.0.0/cessna.osg
>> failed to load
>
> The log entry "Opened DynamicLibrary…" comes from
> DynamicLibrary::DynamicLibrary(name, handle), which shouldn't be reachable if
> the library doesn't load.
> The first "LOADED" log entry comes from Registry::loadLibrary(filename), and
> changes to "PREVIOUSLY_LOADED" the second time I try to load a file.
> The second "LOADED" log entry comes from Registry::read(readfunctor)
>
>> // now look for a plug-in to load the file.
>> std::string libraryName =
>> createLibraryNameForFile(readFunctor._filename);
>> if (loadLibrary(libraryName)!=NOT_LOADED)
>> {
>> printf(" LOADED\r");
>> for(;itr.valid();++itr)
>> {
>> printf(" loop, reading\r");
>> ReaderWriter::ReadResult rr = readFunctor.doRead(*itr);
>> if (readFunctor.isValid(rr)) return rr;
>> else results.push_back(rr);
>> }
>> }
>> else
>> {
>> printf(" NOT_LOADED\r");
>> }
>
>
> The "Warning: Could not find plugin…" is a misnomer - The warning is
> generated if the model couldn't be read, regardless of why.
>
> Here's the interesting bit though - you'll notice that I added "printf("
> loop, reading\r");" inside the read loop, and it's never being reached. I
> have no idea why, except that itr.valid must be returning false.
>
> Any ideas?
>
> TIA,
>
> Ron Aldrich
>
> _______________________________________________
> osg-users mailing list
> [email protected]
> http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
_______________________________________________
osg-users mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
