timothy
fba5233caf
168 records -> docs/decisions/records/<area>/<topic>.md (163 active, 23 dirs) and docs/decisions/archive/<area>/<topic>.md (5 archived). The filename IS the key, so one-active-record-per-key becomes a filesystem property rather than a validator check, and supersession becomes a `git mv`. WHY: the monolith was a concurrency problem before an aesthetic one. A 3,900-line append target made parallel sessions collide -- PR #605 and PR #614 both hit append-vs-append conflicts during routine rebases, and hand-resolving those inside the corpus is exactly the operation the rationale-rewrite guard exists to police. HOW IT IS VERIFIED: a ~170-file diff cannot be meaningfully read, so correctness does not rest on reading it. The parser was taught BOTH formats first, so the body-diff guard parses the old form at the merge-base and the new form at head -- the migration validates itself, no bypass. The proof is a field-level equivalence harness: 168 records before and after, zero lost, zero gained, zero field mismatches, zero rationale bodies differing. Reviewers should scrutinise the harness; it is the actual evidence. What measuring caught that reading would not have: - ~500 lines sit OUTSIDE any record -- decisions.md's lifecycle schema and each topic file's preamble, mostly the only copy. Source files are kept and stripped, never deleted. They also cannot be filed per-area: topic files hold several areas and 4 of 23 areas span several files. - Archive discovery was a non-recursive glob; after the split it found ZERO archived records, surfacing as four bogus "supersedes points to unknown key" errors rather than an obvious failure. - ~32 live docs point into the corpus BY DATE, which the split dangles. Each stripped file now ends with a generated "Records formerly in this file" index, which also rescues the identical breadcrumbs in old issue comments. - decisions.md's "In this file:" list was 97 same-file anchor bullets that the split makes WRONG, not merely stale. Dropped; the generated index replaces them with links that resolve. The equivalence harness now runs against a checked-in FIXTURE, not the live corpus. The earlier version migrated the real tree, which made it a one-shot: the moment the migration landed there was nothing left to move and the tests failed for reasons unrelated to the code. A fixture keeps them testing the SCRIPT rather than the repo's current state. Keys preserved verbatim, warts included: `sched` (12) and `scheduling` (1) remain two directories for one concept. Renaming a key is not a move -- it changes identity, breaks the equivalence proof, and invalidates MemPalace's per-key drawers. Taxonomy normalisation is separate work. refs #610
3.4 KiB
key, title, status, since, supersedes, superseded-by, rule, signals, mechanics
| key | title | status | since | supersedes | superseded-by | rule | signals | mechanics |
|---|---|---|---|---|---|---|---|---|
| concurrency.idempotent-concurrent-add | 2026-07-18 — Concurrent same-item add is idempotent, not a 500: catch the unique-violation per provider (#308) | active | 2026-07-18 | none | none | A concurrent duplicate `Add*ToCollection` that loses the race on the composite-key unique constraint is treated as an idempotent no-op (skip the reindex/rebuild fan-out), not a 500 — detected via a provider-specific `TvContext.IsUniqueConstraintViolation` delegate defaulting to "no". | idempotent add, unique-constraint violation, provider error classifier · paths: `TvContext.IsUniqueConstraintViolation`, `SqliteErrorClassifier`, `MySqlErrorClassifier` · issues: #308, #269, #253 | `docs/api-conventions.md` §7a ("Idempotent insert under concurrency"); `ConcurrencyExtensions.TrySaveChangesForcingVersion` |
Decision. The Add*ToCollection family's membership pre-check (#269) is not atomic with the insert, so two
concurrent adds of the same item both observe it absent and both stage the CollectionItem composite key; the
loser's SaveChangesForcingVersion threw a unique/PK-violation DbUpdateException (SQLite error 19 / MySQL 1062)
it did not catch → 500. We now treat that loss as an idempotent no-op, not an error: the desired end state
(the item is a member) already holds because the racing winner inserted it, rotated the ETag, and fanned out the
rebuild.
Mechanism. A bool-returning sibling ConcurrencyExtensions.TrySaveChangesForcingVersion wraps
SaveChangesForcingVersion and catches only a classified unique/PK violation, returning false. The 10
single-item handlers return Unit.Default on false (skip the reindex/rebuild fan-out — the winner did it). The
bulk AddItemsToCollection handler cannot no-op — that would silently drop the non-colliding items when a batch
partially overlaps a concurrent add — so it retries on a fresh context against recomputed membership (bounded
loop; the common no-collision path runs once).
Provider seam. Detection is provider-specific but the Application layer must not reference the provider
packages, so it follows the existing TvContext static-provider-config idiom (IsSqlite, LastInsertedRowId): a
settable TvContext.IsUniqueConstraintViolation delegate, pointed at SqliteErrorClassifier (extended codes 1555
PK / 2067 UNIQUE) or MySqlErrorClassifier (Number == 1062) from Startup.cs, defaulting to a conservative
"no" so an unwired provider never silently swallows a save failure. Chosen over DI to avoid threading a new
service through 11 handlers, and because the provider discriminator already lives as a TvContext static.
Scope boundary. Add*ToPlaylist is deliberately untouched: PlaylistItem has its own identity PK and no
unique index on (PlaylistId, MediaItemId) — a playlist may legitimately contain the same item more than once, so
there is no constraint to violate.
Tests. A negative-control anchor proves the race genuinely throws a classified DbUpdateException; the fix's
end-to-end handler tests reproduce a real cross-connection race via a shared-cache SQLite harness + a
SavingChanges interceptor that inserts the conflicting row on another connection mid-save (the single-connection
in-memory fixture cannot). Every fix-dependent test was verified to fail with the catch disabled. Mechanics:
api-conventions.md §7a ("Idempotent insert under concurrency"). Refs #308 #269 #253.