[sirikata/sirikata] 6d57ee: Fix waiting count stat in aggregate manager.

[GitHub]

Branch: refs/heads/master
Home: https://github.com/sirikata/sirikata
Commit: 6d57ee96864d1672513b30081a8cecb40b329cda
https://github.com/sirikata/sirikata/commit/6d57ee96864d1672513b30081a8cecb40b329cda
Author: Ewen Cheslack-Postava <ewe...@cs.stanford.edu>
Date: 2013-05-23 (Thu, 23 May 2013)

Changed paths:
M libspace/plugins/mesh/MeshAggregateManager.cpp

Log Message:
-----------
Fix waiting count stat in aggregate manager.

Only count mDirtyAggregateObjects.size() when computing the number of
waiting aggregates since the objects in the mObjectsByPriority lists
that were also being counted should still be in mDirtyAggregateObjects.


Commit: 885f88cb911d0c6f3ad0a4cb7fa4825733ca586d
https://github.com/sirikata/sirikata/commit/885f88cb911d0c6f3ad0a4cb7fa4825733ca586d
Author: Ewen Cheslack-Postava <ewe...@cs.stanford.edu>
Date: 2013-05-23 (Thu, 23 May 2013)

Changed paths:
M libspace/plugins/mesh/MeshAggregateManager.cpp
M libspace/plugins/mesh/MeshAggregateManager.hpp

Log Message:
-----------
Fix queuing of dirty aggregates and stats.

Only increase the number of queued aggregates if the aggregate wasn't
already in the queue. Not doing this was leading to an ever-increasing
queued count even though new objects weren't really being added.

Also track uploading aggregate meshes because the two existing
containers can't distinguish between the four possible states (dirty
and unqueued, queued for generation, generated and uploading, or
finished uploading). Use this to avoid re-enqueuing a mesh that's
still being uploaded.


Compare: https://github.com/sirikata/sirikata/compare/882babf7f6b7...885f88cb911d