Fixes an issue where long inputs could cause issues when dealing with formspecs.
The expected data is usually below 1 KiB, however that's not a technical limit.
* Reset usage timers in loops culler
* Correctly touch map blocks when using bfs culler
* Align use of variables
* Report statistics when using loop culler
* Restart map-send cycle if visible blocks are soon to be unloaded.
* Reset completion timer when camera or position changed.
Co-authored-by: sfan5 <sfan5@live.de>
In very narrow windows, `m_cols` can be small (i.e. 0).
Hence, `m_view <= m_line.size() + 1 - m_cols` does not guarantee
`m_view <= m_line.size()`.
`std::string::substr(pos, npos)` requires `pos <= size()`.
Cache blocks already occluded at a specific distance. The RemoteClient typically visits the same distance multiple time - especially at larger distances, so this saves significant CPU from recalculating the occlusion state of blocks.
This obsoletes the current client-side setting entirely. The server can
transmit the tilt to the client directly and will send 0.0f as default
value.
Co-authored-by: x2048 <codeforsmile@gmail.com>
Co-authored-by: sfan5 <sfan5@live.de>
When mesh chunking and shadows are enabled, make sure that the mesh-holding blocks are added to the shadow drawlist.
Otherwise those portions of the shadows will not be rendered.
* Generalize mesh chunking. Set 3x3x3 chunks.
* Make mesh chunk size configurable... Default to 1 (off).
* Extract all mesh grid maths into a dedicated class
---------
Co-authored-by: x2048 <codeforsmile@gmail.com>
Reduce the number of drawcalls by generating a mesh per 8 blocks (2x2x2). Only blocks with even coordinates (lowest bit set to 0) will get a mesh.
Note: This also removes the old 'loops' algorithm for building the draw list, because it produces visual artifacts and cannot be made compatible with the approach of having a mesh for every 8th block without hurting performance.
Co-authored-by: Jude Melton-Houghton <jwmhjwmh@gmail.com>
Co-authored-by: Lars <larsh@apache.org>
Co-authored-by: sfan5 <sfan5@live.de>
