Uh oh!
There was an error while loading. Please reload this page.
fix(cli): extend better-sqlite3 → wasm SQLite auto-fallback to the persistent-file / --artifact dev path - #2234
Merged
Conversation
…rsistent-file / --artifact dev path (#2229) The native-better-sqlite3 → wasm SQLite → in-memory step-down previously only guarded the zero-config `:memory:` dev branch of `serve`. A normal `objectstack dev` run never reached it: `dev` injects a persistent `file:` DB so AI-authored data survives restarts, and `--artifact` boots resolve sqlite through the datasource factory — both constructed `better-sqlite3` directly with no probe and no fallback. better-sqlite3 loads its native addon lazily (first query), so an ABI mismatch was invisible at boot and surfaced later as a runtime `Find operation failed`. Hoist the probe-by-connect + step-down into a shared `resolveSqliteDriver` helper (@objectstack/service-datasource) and apply it to both previously unguarded sqlite construction sites: the explicit `sqlite`/`file:` branch in serve.ts and the sqlite branch of the default datasource driver factory. The helper forces the native load with `connect()` + `SELECT 1` and, in dev only, steps down to wasm SQLite (real SQL + on-disk persistence — the same `file:` keeps working) then to in-memory as a last resort, emitting the existing `⚠ native better-sqlite3 unavailable …` warning. In production the native driver is returned unprobed so a load failure surfaces loudly (fail-closed) instead of silently degrading. Dev is plumbed through standalone-stack so the --artifact boot enables the fallback explicitly (NODE_ENV fallback otherwise). Adds a co-located unit test simulating a NODE_MODULE_VERSION load failure (asserts wasm fallback + warning, in-memory last resort, and prod fail-closed). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The latest updates on your projects. Learn more about Vercel for GitHub.
|
Contributor
📓 Docs Drift CheckThis PR changes 3 package(s): 29 hand-written doc(s) reference the affected code and may need an implementation-accuracy re-verification:
|
Uh oh!
There was an error while loading. Please reload this page.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for freeto join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closes#2229.
Problem
The native-
better-sqlite3→ wasm SQLite → in-memory auto-fallback only existed in the zero-config:memory:branch ofserve(theelse if (isDev)default-dev branch). A normalobjectstack devrun never reaches it:devinjects a persistentfile:DB by design (so AI-authored metadata/records survive restarts) → routes into the explicitsqlite/file:branch inserve.ts, which didnew SqlDriver({ client: 'better-sqlite3', … })directly — no probe, no fallback.--artifactboots resolve sqlite throughdefault-datasource-driver-factory.ts(viastandalone-stack) — alsonew SqlDriver({ client: 'better-sqlite3', … })with no probe/fallback.better-sqlite3loads its native.nodeaddon lazily (at the first pool-connection acquire / first query, not atrequireor knex construction). So an ABI mismatch (e.g. a cached prebuilt binary built for a different Node version —NODE_MODULE_VERSIONmismatch) was invisible at boot and surfaced later as a runtimeFind operation failedon the first query.Fix
Hoist the probe-by-connect + step-down into a shared
resolveSqliteDriverhelper in@objectstack/service-datasourceand apply it to both previously-unguarded sqlite construction sites:sqlite/file:branch inserve.ts, andThe existing default-dev
:memory:branch is refactored onto the same helper, so there is now a single implementation.The helper:
connect()+ a cheapSELECT 1(connect()alone is not a reliable probe — for SQLite it only doesmkdir+ aPRAGMAwhose error it swallows; the rawSELECT 1propagates the load error).file:keeps working) preserving thefilename, then toInMemoryDriveras a dev-only last resort, emitting the existing warning verbatim:StandaloneStackConfig/serve.ts(with aNODE_ENV==='development'fallback for programmatic callers).Tests / verification
packages/services/service-datasource/src/sqlite-driver-fallback.test.ts(8 cases) simulates aNODE_MODULE_VERSION-style load failure and asserts: wasm fallback + the exact warning,on-write/on-disconnectpersistence selection, in-memory last resort,autoMigrate/schemaModepass-through, prod fail-closed (native returned, never degraded), and theNODE_ENVdev-gate default.pnpm test --filter @objectstack/service-datasource→ 93 passed.turbo run buildfor@objectstack/service-datasource,@objectstack/runtime, and@objectstack/cliall green (DTS typecheck included).resolveSqliteDriveron:memory:, a persistentfile:(with a realINSERT/SELECTround-trip), anddev: falseall resolve to nativebetter-sqlite3with no warning — confirming no behavior change on the happy path.Changeset
patchfor@objectstack/service-datasource,@objectstack/cli,@objectstack/runtime.🤖 Generated with Claude Code