Feature Requests

The shape that detects which avatars to prioritize rendering, seems to be roughly spherical and based on the FOV of the avatar somehow. As the camera is not spherical, it seems odd to choose a spherical algorithm for this, especially since half of that view should be unrendered. Why does this matter? It's common that only 1/3 of visible avatars are actually rendered as non-impostors. The shape used for non-imposter calculation is putting several avatars, who cannot be seen, as fully rendered. This means 2/3 of my avatar rendering capability is being wasted on things I cant see at all!

Please increase the server notecard cache size. The server notecard cache size is 3 MiB, which is tiny by any standard. At the advent of llGetNotecardLineSync(), which completely relies on cache integrity, it has been shown that some regions exhaust/cycle the cache more frequently than others. In some cases, the cache is cycled so quick that not even one line can read via llGetNotecardLineSync() after the notecard is cached via calls from llGetNumberOfNotecardLines() or llGetNotecardLine(). I propose an increase to at least 64 MiB. This should maintain at least 1000 max size notecards and not 45 which the current size maintains. This will also significantly reduce dataserver queries from regions that have constant cache cycling. Thanks.