Gui object namespace issue

#1 Haven't tested your code but I got this error whenever I tried to assign a Name to a SkeletonObject(in my Spine branch).

%obj.setName("test");

->throws an error upon toy reload/restart/quit

No errors without that.

#2 Is that object assigned a script class? I'm thinking something has broken in the namespace linking system in all of the clean up. I'm slowly working through file comparisons now to see if I can track it down.

#3 Hack fix: mRefCountToParent >= 0. Others have found different solutions.

#4 Guess it works in the short term - I'll keep looking though. This needs to work correctly. When I get it straightened out I'll drop a pull request.

Thanks!

#5 Ok, Master branch has this:

bool Namespace::unlinkClass(Namespace *parent)
{
   Namespace *walk = this;
   while(walk->mParent && walk->mParent->mName == mName)
      walk = walk->mParent;

   if(walk->mParent && walk->mParent != parent)
   {
      Con::errorf(ConsoleLogEntry::General, "Namespace::unlinkClass - cannot unlink namespace parent linkage for %s for %s.",
         walk->mName, walk->mParent->mName);
      return false;
   }

   mRefCountToParent--;
   AssertFatal(mRefCountToParent >= 0, "Namespace::unlinkClass - reference count to parent is less than 0");

   if(mRefCountToParent == 0)
      walk->mParent = NULL;

   trashCache();

   return true;
}

Development branch has this:

bool Namespace::unlinkClass(Namespace *parent)
{
   Namespace *walk = this;
   while(walk->mParent && walk->mParent->mName == mName)
      walk = walk->mParent;

   if(walk->mParent && walk->mParent != parent)
   {
      Con::errorf(ConsoleLogEntry::General, "Namespace::unlinkClass - cannot unlink namespace parent linkage for %s for %s.",
         walk->mName, walk->mParent->mName);
      return false;
   }

   AssertFatal(mRefCountToParent > 0, "Namespace::unlinkClass - reference count to parent is less than 0");
   mRefCountToParent--;

   if(mRefCountToParent == 0)
      walk->mParent = NULL;

   trashCache();

   return true;
}

Master is generally in line with what T3D does, though the particulars are different. Was there a reason for this particular change? Is there something we're trying to gain that warrants the additional research that is necessary to make this work correctly?

What's really funny is that they look like they do the same thing. <shrug>

#6 Here's the background on this change:

github.com/GarageGames/Torque2D/pull/126

#7 <edit>
I take it back - that fix is already applied (commented not removed) and this assert still gets triggered in Development, whereas the Master version continues to be stable. I guess I'll dig a little more....

#8 Looks like the problem is the taml parser. It's calling setClassNamespace and setSuperClassNamespace, which assume that the object has already been added.

Think this works

void SimObject::setClassNamespace( const char *classNamespace )
{
   mClassName = StringTable->insert( classNamespace );
   if (mFlags.test(Added))
       linkNamespaces();
}

void SimObject::setSuperClassNamespace( const char *superClassNamespace )
{  
   mSuperClassName = StringTable->insert( superClassNamespace );
    if (mFlags.test(Added))
        linkNamespaces();
}

#9 Very nice! Thanks tons - hadn't gotten that far yet.

#10 Sent in a pull request for this.