max / makenotwork

server: correct enum VARIANTS doc to match drift-test coverage The VARIANTS doc no longer claims the enum-drift test covers every backing column; coverage is the test's explicit registry, not every enum automatically. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: Max Johnson <me@maxj.phd> · 2026-07-05 15:50 UTC

Commit: 8d6f457628f68e22fe85b899b26401defae9ad49

Parent: f58201d

1 file changed, +9 insertions, -6 deletions

M server/src/db/enums.rs +9 -6

			@@ -82,12 +82,15 @@ macro_rules! impl_str_enum {
82	82		}
83	83
84	84		impl $enum_name {
85		-	/// Every wire/DB string this enum maps to. The single source of truth
86		-	/// for the variant set; the enum-drift test asserts it equals the
87		-	/// Postgres `CHECK (... IN (...))` list for each backing column, so a
88		-	/// variant added here without widening the DB constraint (or vice
89		-	/// versa) fails at test time instead of at the first read of a
90		-	/// poisoned row.
	85	+	/// Every wire/DB string this enum maps to — the single source of
	86	+	/// truth for the variant set. For each enum registered in the
	87	+	/// enum-drift integration test (`tests/workflows/enum_drift.rs`),
	88	+	/// this set is asserted equal to the Postgres `CHECK (... IN (...))`
	89	+	/// list on its backing column, so a variant added here without
	90	+	/// widening the DB constraint (or vice versa) fails at test time
	91	+	/// rather than at the first read of a poisoned row. Coverage is that
	92	+	/// registry, not every enum automatically: add a `(enum, table,
	93	+	/// column)` row there when a new CHECK-constrained column lands.
91	94		pub const VARIANTS: &'static [&'static str] = &[$($str),+];
92	95		}
93	96		};