Glean: Data about Code
summary
Glean is an open-source system that indexes code into strongly typed facts stored in RocksDB and answers millisecond queries via a Datalog-like language called angle; indexers use native compiler front ends (Clang for C++) and a command-line glean index/interactive shell workflows. It runs at scale (example: indexing React with Flow produces 430,000 facts in 15 MB), compiles queries to bytecode, has a Haskell type-safe DSL and codegen, and uses the haxel/applicative-do stack to run many queries concurrently.
AI-written summary · about
chapters (8)
- 0:00Introduction to Glean
- 2:24System architecture
- 4:43Data model and schemas
- 9:27Querying with Angle
- 12:20Language-agnostic querying
- 17:25Code markup layer
- 20:29Haskell-based DSLs
- 27:39Concurrency and availability
related talks
01h 22mGlean: C++ Code Indexing at Meta
10mRefactoring React code with Glean
19mJumping on the Bandwagon: How I ended up writing Gleam for a living
36mCode as Data: How data insights on legacy codebases fill the knowledge gap in modernization projects
04mGet Productive on Unfamiliar Source Code
20mAdopting Gleam the Boring Way
20mDevX: History of Code Search
38mRapidly building with Gleam