CQL conformance test infrastructure using upstream ScyllaDB/Cassandra tests

[Copilot]

Run 313 existing CQL tests from ScyllaDB's test/cqlpy/cassandra_tests/ (ported from Apache Cassandra) against objstore to track protocol implementation correctness.

Native protocol fixes for cassandra-driver compatibility

• Add COMPRESSION key to OPTIONS/SUPPORTED response (Python driver requires it)
• Process all buffered frames per TCP read, not just one — fixes stream multiplexing timeouts when drivers send parallel metadata queries
• Return empty result sets for unimplemented system_schema tables (indexes, triggers, types, functions, aggregates, views, vertices, edges) so driver metadata sync completes

Test runner (extra/cql_tests/)

Fetches test content from upstream ScyllaDB at a configurable git ref, starts objstore, runs via pytest:

# Run all 313 tests
bash extra/cql_tests/run.sh

# Filter to specific tests
bash extra/cql_tests/run.sh -t "insert_test or select_test"

# List available tests
bash extra/cql_tests/run.sh -l

Compatibility shims (conftest.py, util.py, nodetool.py) provide the fixtures and helpers the upstream tests expect — no test content is written or maintained locally.

GitHub Action (cql-conformance.yml)

Builds objstore, fetches tests, runs suite. Supports workflow_dispatch with test_filter (pytest -k expression) and scylla_ref inputs.

Original prompt

---

This section details on the original issue you should resolve

<issue_title>Research test content for driving CQL implementation correctness</issue_title>
<issue_description>Research by looking at the scylladb and apache cassandra repositories and documentation for existing end-to-end test content using the CQL native protocol. The test should ideally be focused and as minimal as possible, but the aim is to have as much test content as possible. Draft implementing a github action which allows running the test suite and subsets of the test suite, along with tracking overall progress. </issue_description>

Comments on the Issue (you are @copilot in this section)

• Fixes Research test content for driving CQL implementation correctness #26

---

💬 Send tasks to Copilot coding agent from Slack and Teams to turn conversations into code. Copilot posts an update in your thread when it's finished.