llama.cpp Release b10935 Adds LOG_JSON Macro for Structured Logging
llama.cpp release b10935 introduces a new `LOG_JSON` macro within the common utilities to support structured data logging. The release also includes a demo implementation demonstrating how to log JSON data. Structured JSON logging enables automated log parsing and telemetry collection for observability and monitoring systems. This makes it easier for developers integrating llama.cpp into enterprise pipelines to process and analyze runtime metrics. The update centers on PR #28586, which adds the `LOG_JSON` macro to the common codebase. Pre-built binaries were released for multiple operating systems and backends, including Windows CUDA/Vulkan/ROCm, Linux, macOS, Android, and iOS.
## BACKGROUND
llama.cpp is a popular open-source inference engine designed for running Large Language Models efficiently on consumer hardware using C and C++. Structured logging converts unstructured text logs into standardized formats like JSON, making them machine-readable for downstream monitoring tools.