Ok now we are getting deep into the nuts and bolts, RC2 build 2145 -Lets explore the focus of this exercise - More UI work, Message Template matches the latest LL version.Python code fixes, Fix to panel snapshot from 2025.4 , Fix an edge case where TGA's fail to upload, related to image preview. Auto and Nullptr used where applicable. More UI and extensions to low level tweaks. Local parcel Twins fix. Add OpenXR to the mix. Reduce stats spam to console, SSR mods and removal of redundant 3D code. Increase randomness in shaders relating to reflection from 128 vectors to 256 and add jitter to the shader. Unroll some of the most hideous nested loops, and refactor prim code both for cubes , pyramids and physics around spheres. exploit opportunities to parallel code in this area. improve the client packer non LL version. More AVX for v3math. And finally completly rewrite the entire perlin math code, for noise , turbulence and clouds.
Thats a wall of text :) now we are back at the edge, I seem to be revisiting code which has had little to no attention for the last 10 years, well it works... but does it work well? The focus sometimes is more on new features or additions which may be handy for some but essentially do little for the core experience.
The problem is new stuff is fantastic, but at what cost to the overall end product? And in the cold hard light of day, why are loops nested 5 deep, why have warnings for altivec processors? Devs know the bit im talking about :) In the persuit of progress are we losing sight of making what we have work as well as it can?
Im currently testing that theory!
Anyway, enjoy the weekend,
Love KL
Comments
Post a Comment