Table of Contents

How to manage the example database

Conjecture can persist failing test inputs in a local SQLite database, ensuring known failures are always re-checked on subsequent runs.

Note

For background on why the database exists and what it stores internally, see Understanding the example database.

How it works

  1. A [Property] test fails with some input
  2. Conjecture saves the failing byte buffer to the database
  3. On the next run, stored examples are replayed before generating new random inputs
  4. If the test now passes, the example stays as a permanent regression test
  5. Future runs continue to replay it

Configure the database path

The default path is .conjecture/examples/. Override it at the assembly or test level:

// Assembly-level
[assembly: ConjectureSettings(DatabasePath = ".test-data/conjecture/")]

// Per-test
[Property(UseDatabase = true)]
public bool My_property(int value) => ...;

Disable the database

Globally — for CI or stateless test runs:

[assembly: ConjectureSettings(UseDatabase = false)]

Per-test:

[Property(UseDatabase = false)]
public bool Stateless_test(int value) => ...;

CI only:

#if CI
[assembly: ConjectureSettings(UseDatabase = false)]
#endif

Version control options

Option A: Commit the database

Add .conjecture/examples/ to your repo:

  • CI replays the same known failures
  • Team members share discovered edge cases
  • Failures become permanent regression tests

Option B: Ignore the database

Add to .gitignore:

.conjecture/

Each developer and CI run starts fresh. Good for projects that want stateless, independent test runs.

Option C: CI-only disable

Keep the database locally, disable in CI via an environment variable and a conditional directive (see above).

Settings reference

Setting Type Default Description
UseDatabase bool true Enable/disable database persistence
DatabasePath string ".conjecture/examples/" Directory for database files

See Reference: Settings for the full settings table.

Edit this page