Hi Chris,

I'll hold back from doing any more review until you've submitted the
other work that depends upon it.  I still am of the view that it's
probably not the right way to tackle the problem in hand.

>  I don't believe the ReadFileCallback is a safe way to integrate with the 
> pseudoloader.

Could you explain why you don't think it's a safe way to integrate
with a pseudoload?

Robert.


On Thu, Nov 19, 2009 at 2:55 PM, Chris 'Xenon' Hanson
<[email protected]> wrote:
> Robert Osfield wrote:
>> Hi Chris,
>> I have just reviewed the your suffix change and I'm not sure I quite
>> get why you need it.
>
>  This code is to support additional functionality that I haven't yet been 
> permitted to
> release, but I expect to soon.
>
>> The PagedLOD's DatabasePath member is required to make sure that path
>> that a tile is loaded from at runtime it recorded so that it's
>> children can be found, this DatabasePath is set on load of the node so
>> is actively used.
>> This contrasts your new DatabaseSuffix change, it's not recorded or
>> set anywhere in the OSG, so I can only assume it's your app that is
>> setting, and if so then why not just append the suffix to the
>> childrens filenames directly?  Also if this appending of the suffix is
>> something done dynamically then the route I would take is to write a
>> custom Registry::ReadFileCallback that intercepts all readNodeFile()
>> etc calls and then adds the suffix there.  Such a callback avoids the
>> need to modify the scene graph at all so will be easier to implement
>> and more more efficient.
>> Given that I feel that your changes tackle a problem in way that I
>> would recommend I don't feel it's appropriate to merge them as they
>> just add complexity to the implementation for little or no value add,
>> and in fact may just confuse users into tackle things in a less
>> optimal way.
>
>  The non-persistent suffix is there to support propagation of a pseudoloader 
> suffix
> throughout the process of loading a PagedLOD database, without persistently 
> altering the
> structure or filenames of the resulting database once it is loaded. This way 
> it can be
> re-saved with the modifications caused by the pseudo-loader, but without any 
> permanent
> alteration to the filename structure of the database. This is needed because 
> if you apply
> a pseudoloader to the top-level PagedLOD scenegraph file, the filenames 
> embedded in the
> PagedLOD nodes' perRangeLists don't automatically inherit this.
>
>  I have a visitor that appplies this suffix to the PagedLODs automatically, 
> but it needs
> to do it without permanently tainting the perRangeList child filenames -- 
> thus the
> emphemeral suffix which isn't saved.
>
>  You and I discussed the terrain-modification approach this summer and the 
> pseudoloader
> technique was your own suggestion.
>
>  I don't believe the ReadFileCallback is a safe way to integrate with the 
> pseudoloader.
>
>> Robert.
>
> --
> Chris 'Xenon' Hanson, omo sanza lettere                  Xenon AlphaPixel.com
> PixelSense Landsat processing now available! http://www.alphapixel.com/demos/
> "There is no Truth. There is only Perception. To Perceive is to Exist." - Xen
> _______________________________________________
> 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