Databricks has open-sourced Metals v2, a language server for Scala and Java built to handle codebases with millions of lines. The system uses three architectural layers: a build-free repository index that enables immediate code intelligence, compiler-backed pipelines for Scala and Java that use source-path based symbol loading, and metadata-first build integration that queries Bazel targets at editor speeds. The Scala pipeline maintains a single presentation-compiler instance managing 24 million lines of code while delivering diagnostics at a median latency of 0.9 seconds, while the Java implementation uses a modified Turbine header compiler that processes roughly one million lines per second to stay interactive.
