2 Commits (6d6015a2034dbee2ea4eb50d2e84381df008fa46)

Author	SHA1	Message	Date
René Cannaò	bcba8039cd	Address review feedback on query cache unit tests (PR #5486) ... - test_pgsql_flush: flush() at start for clean state, assert == 10 - test_mysql_construction_and_flush: fix tautology (flushed == 0 || flushed >= 0 is always true for uint64_t), double-flush to ensure empty cache, then assert == 0 - test_pgsql_purge_expired: use max_memory_size=1 instead of 1GB to force the purge path to actually run (3% threshold wasn't reached) - test_pgsql_many_entries: fix misleading docstring claiming hash table distribution verification	3 weeks ago
René Cannaò	96387be177	Add Query Cache unit tests (Phase 2.3, #5475) ... Comprehensive unit tests for MySQL_Query_Cache and PgSQL_Query_Cache covering 26 test cases across 11 test functions. Runs in <0.01s with no infrastructure dependencies. PgSQL_Query_Cache is used for most tests because its set()/get() API is simpler (no MySQL protocol parsing required). The underlying Query_Cache<T> template logic is identical for both protocols. Test coverage: - Basic set/get cycle with value verification - Cache miss on nonexistent key - User hash isolation (same key, different user_hash = miss) - Cache replacement (same key overwrites previous value) - Hard TTL expiration (expire_ms in the past) - Soft TTL expiration (cache_ttl parameter to get()) - flush() clears all entries and returns correct count - Set/flush cycle (entry retrievable then unretrievable) - Stats counters (SET and GET via SQL3_getStats()) - SQL3_getStats() result format (2 columns, has rows) - MySQL_Query_Cache construction, flush, and SQL3_getStats() - purgeHash() eviction of expired entries (live entries survive) - Bulk insert (100 entries) across 32 internal hash tables Key implementation note: Query_Cache constructor requires a valid Prometheus registry (GloVars.prometheus_registry). The test_init_ query_cache() helper creates one automatically.	3 weeks ago