Quantified API

V1.4.3 - New Multiloader Architecture & End of Life for Legacy Versions :L It's sad to say farewell to Forge 1.18.2 and Forge 1.19.2. They served their time, but the time has come to let them nap. Both versions are officially now EOL and will no longer receive updates or support in any way.

--- What's changed ---

Architecture:

• Complete rebuild of the project structure around a new multiloader design (peak). All platform implementations (Forge 1.20.1, Forge 1.21.1, Fabric 1.20.1, Fabric 1.21.1, NeoForge 1.21.1) now live under platforms/ as subprojects, sharing a single core/ source set (one jar, all loaders)
• Universal jars now built per-version (universal-1.20.1, universal-1.21.1) plus a combined omniversal jar covering all supported loaders and versions.
• Gradle build orchestration fully centralised in the root build.gradle with dedicated GradleBuild tasks per platform.

Platforms dropped:

• forge-1.18.2 (EOL)
• forge-1.19.2 (EOL)

Old/New Platforms:

• platforms/forge-1.20.1
• platforms/forge-1.21.1
• platforms/fabric-1.20.1
• platforms/fabric-1.21.1
• platforms/neoforge-1.21.1

Description changes:

• Updated README and added a new CFREADME file for Curseforge Description.
• Updated the Curseforge Description.

Core fixes (since V1.4.2-HOTFIX):

• VulkanBatchWorkload, GpuTaskDispatcher: async dispatch improvements.
• AutoHealthChecker, MemoryCore, DeveloperDashboardServer: internal cleanup.
• ConnectedModImpl: compatibility fixes.
• New McDensityProgram and McDensityVulkanTask added to core.
• New platform abstraction layer under core/common/platform/.
• Fixed Vulkan GPU acceleration selecting an worser GPU by default, due to higher VRAM size.
• Fixed an issue where switching between preferred Vulkan Devices doesn't reprobe the newly selected GPU and falls back to OpenCL.