CQLite v0.12.0

CQLite v0.12.0 — the compaction release

Read, query, and now compact Apache Cassandra 5.0 SSTables — no cluster required. This is our biggest release yet.

✈️ Arrow Flight + Trino connector

• Run SQL over your SSTables from Trino — a built-in Arrow Flight server exposes raw Cassandra data files as a federated source, no Cassandra cluster in the loop.
• Pushdown that prunes: predicate pushdown, token-range SSTable pruning (skip files that can’t match), and nested predicates (OR/NOT).
• Aggregation pushdown — count/sum/min/max/avg + GROUP BY evaluated close to the data, with an operator gate for high-cardinality groupings.

📤 CDC delta-export to Parquet

• Stream SSTable changes straight to Parquet — cqlite delta-export projects an SSTable generation into a CDC-style delta envelope: upserts plus row/range/partition tombstones with per-cell metadata.
• Drop it into your lakehouse — feed the envelopes into DuckDB, Spark, or any Parquet consumer; ships with a DuckDB reference reconciliation guide for correct latest-state without a running cluster.
• Embeddable Parquet export from Python and Node (export_parquet, configurable row-group size + compression).

🧱 Byte-for-byte compaction parity

• cqlite compact produces SSTables identical to Apache Cassandra’s — verified in CI by a differential harness and an end-to-end Cassandra readback. Full reconciliation semantics: complex and range tombstones, gc_grace purging, TTL tie-breaks, and multi-cell collections and UDTs.

Plus

• Canonical BTI (da) support — read and write Cassandra’s trie-indexed format.
• WRITETIME() and TTL() in SELECT for debugging and migration.
• Richer query engine — PER PARTITION LIMIT, static columns, clustering order and bounds, and partition-targeted lookups that skip full scans.
• Faster reads — a size-aware disk backend with direct I/O (O_DIRECT/F_NOCACHE) for huge files, so a big scan won’t thrash the page cache.

Get it

brew install pmcfadin/cqlite/cqlite      # CLI (macOS + Linux)
cargo add cqlite-core                     # Rust library
pip install cqlite-py                     # Python
npm install @cqlite/node                  # Node.js

Full detail in the CHANGELOG.