Hi All,

I have just checked into svn/trunk a re-factor of DatabasePager that
hopefully will address some of the outstanding issues with use of paged
databases that users which deleted subgraphs independantly from the
DatabasePager, or for paged databases where subgraphs are cached
independently from the DatabasePager.  In both these cases it was possible
to create subgraphs that were effectively orphaned and never cleaned up
properly.  My hope that this issue will now be fixed and users who wish to
use file caching in conjunction with paged databases will be free to do so.
There have been reports of clean up issues with the txp database which again
todays changes should help clear up (in conjunction with some other recent
changes checked into txp plugin.)

The re-factor also makes it possible to load new PagedLOD subgraphs and
merge them with the scene graph independently from the DatabasePager without
the need to call DatabasePager::registerPagedLODs(..) on these subgraphs, as
the registerPagedLODs will now happen automatically the first time once any
children of PagedLOD are requested by the cull traversal.   Applications
that missed such a call previously when loading new databases were exposed
to issues of the DatabasePager not knowing about the new subgraphs and
therefor never expiring them.

With any changes there is the potential for regressions, in build across
platforms, performance and stability - I have done lots of testing today but
don't have access to all your applications, data and usage models so please
test the new pager now checked into svn/trunk.  I make this request to
everybody - those who've not been seeing any problems with the pager, and
those who have come up against the above issues.  Please let me know about
how things work out for you, both successes and any remaining issues.

Thanks for your help in testing,
Robert.
_______________________________________________
osg-users mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org

Reply via email to