Faster C++ code intelligence with whole codebase indexing
By Steven Van ·
A persistent symbol index speeds up finding definitions and references in large C++ projects using Copilot CLI.
The Microsoft C++ Language Server used in GitHub Copilot CLI now supports whole codebase indexing (WCI), speeding up C++ code intelligence. Previously, code-intelligence requests could need to rediscover project information as you navigated, slowing down finding a definition, locating references, or understanding unfamiliar code. WCI instead builds a persistent index of symbols across an entire C++ project, including files that aren't currently open, and reuses that index instead of rediscovering the information for each request.
WCI is enabled by default. The language server builds the index the first time a C++ project is opened, which can take extra time and temporarily raise memory usage on large or complex repositories; after that initial build, the index is reused and updated dynamically. Indexing progress can be checked with /lsp logs, and WCI can be temporarily disabled via the indexing documentation (a Copilot session restart is required after changing the setting).
Read the full announcement for details.
