mongo-ruby-driver Validation Report¶
Generated 2026-05-10 — SecantusDB 0.4.0b12 vs mongo-ruby-driver f68d676643c1 (vendor/mongo-ruby-driver/).
Run uv run python -m invoke validate-ruby to refresh. The pass rate is the analogue of the pymongo / mongo-go-driver / mongo-node-driver / mongo-java-driver gauges for the official Ruby driver — the same gem Rails + Sinatra applications and the Ruby ecosystem build on.
Summary by category¶
Category |
Passed |
Failed |
Pending |
Total |
Pass rate |
|---|---|---|---|---|---|
|
71 |
0 |
7 |
78 |
100.0% |
|
26 |
0 |
8 |
34 |
100.0% |
Overall |
97 |
0 |
15 |
112 |
100.0% |
Run time: 2.60s.
How this is generated¶
mongo-ruby-driver’s tests are run unmodified, against a standalone SecantusDB daemon. The submodule at vendor/mongo-ruby-driver/ is checked out at the pinned upstream tag with zero local edits. ruby_validation/runner.py runs bundle install (one-time per checkout), then does a two-phase spawn: phase 1 boots python -m secantus --port 20718 --storage-path <tempdir> without --auth and uses pymongo to createUser root-user (root role) and ruby-test-user (readWriteAnyDatabase + dbAdminAnyDatabase); phase 2 stops that daemon and restarts on the same tempdir with --auth, so the user records persist and the server now enforces auth. rspec runs against phase 2 with MONGODB_URI=mongodb://root-user:password@127.0.0.1:20718/?authSource=admin. On-disk tempdir is rmtree’d after the run.
These are integration specs — every test opens a real TCP connection to the SecantusDB daemon, SCRAM-authenticates, and exchanges wire commands end-to-end. The pass rate is therefore a true measure of SecantusDB’s compatibility with the Ruby driver, not of the driver’s own pure-code logic.
The include set is currently narrow on purpose: a single broken test in spec/mongo/cursor_spec.rb or spec/mongo/bulk_write_spec.rb can pin the runner indefinitely on a tailable getMore that never completes. Each new file is added to include_paths.py only after a manual confirmation that it terminates within the runner’s wall-clock guard (300 s by default).
Pending tests are honest skips driven by mutually-exclusive environment gates upstream tests opt into: require_no_auth (skips when auth is configured — and our gauge always runs with auth), require_topology :single (skips on replica-set topology — we always advertise as a single-node RS primary so change streams work), min_server_version 4.5 (skips on >= 4.6 — we report 7.0.0). Resolving them would require running additional gauge configurations rather than fixing SecantusDB behaviour. The pending number is therefore a lower bound on tests SecantusDB “could pass” if the runner spun up alternate daemons; the failed number is the only signal that matters for conformance.