Skip to content

0.9.25

Choose a tag to compare

View all tags
@cevheri cevheri released this 18 Jun 12:47
v0.9.25
96eab52
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

What's Changed

Schema introspection reliability + correctness

LibreDB Studio 0.9.25 is a reliability-and-correctness release. Its headline is a fix for schema introspection timing out on large PostgreSQL schemas (the table tree could fail to load entirely), backed by 30+ new regression tests and a tightened quality bar (100% coverage on new code). It also rolls up everything since v0.9.19 — SQL identifier-quoting correctness fixes, a Railway one-click template, a Koyeb deploy path, dependency/toolchain upgrades, and a knip dead-code gate.

This is a cumulative release covering 0.9.20 → 0.9.25. v0.9.19 was the previous GitHub Release.

Highlights — schema introspection no longer times out on large schemas (#71)

On schemas with 100+ tables, getSchema() ran a single query joining five information_schema CTEs. Because PostgreSQL 12+ inlines single-reference CTEs and estimates rows=1 for information_schema views, the planner chose nested-loop joins that re-executed the expensive FK/index CTEs many times — ballooning to ~295s and always hitting the connection's 60s statement_timeout. The schema tree never loaded.

Two complementary fixes:

  • AS MATERIALIZED CTEs — each CTE is now computed once instead of being re-executed inside nested loops. ~295s → ~2.6s on a 122-table schema.

  • Two-phase schema loading so a slow/failing stats query can never block the table list:

    • getSchemaList() → tables + columns + PKs (fast, ~50ms) — /api/db/schema/list
    • getSchemaRelations() → foreign keys + indexes (heavy, ~2.3s) — /api/db/schema/relations

    The explorer renders the tree from the fast list immediately, then merges relationships asynchronously. If the relations query fails or times out, it's logged and skipped — the table list stays intact. New optional provider methods fall back to getSchema() / [] for databases that don't implement the fast path, so non-PostgreSQL engines are unaffected.

Database correctness fixes

  • Cross-schema foreign keys no longer dropped (#71) — FK introspection joined constraint_column_usage on table_schema (the referenced table's schema), silently discarding any foreign key that crossed schemas. Now joined on the constraint's own schema.
  • Schema-qualified identifier quoting, per-segment (#70) — schema.table identifiers are quoted segment-by-segment so each part is escaped correctly.
  • Dialect-aware identifier quoting in generated queries, and quoted identifiers in the PostgreSQL table-stats monitoring query.

Testing & code quality

  • 30+ new regression tests locking in the two-phase loading contract: provider methods (getSchemaList/getSchemaRelations, incl. cross-schema prefixing, empty/null edge cases), both new API routes (auth, validation, provider fallbacks, error mapping), and the client hook's resilience guarantee (relations failure must not wipe the table list).
  • 100% coverage on new code and a green SonarCloud quality gate (duplication brought under threshold via shared CTE fragments and a shared route handler).
  • knip dead-code gate added to CI — removed unused dependencies and dead barrel re-exports; configured entry points/scope.
  • Merged-lcov fix — strip non-executable lines so coverage reporting is accurate.

Deployment & distribution

  • Railway one-click template — template.json service definition, logo, per-variable descriptions, and a published Deploy on Railway button.
  • Koyeb — fixed the deploy-button storage default and added a catalog submission guide.

Dependencies & toolchain

  • Dependency refresh with TypeScript 6 adaptation; better-sqlite3 12.10.1; Node.js/Bun pins retained (24.16.0 / 1.3.14); base image node:24.16.0-trixie-slim.

Docs

  • Consolidated overlapping guides and archived shipped plans; added an enterprise-features roadmap and Kubernetes/Helm setup docs; aligned Node/Bun versions across templates and CONTRIBUTING.

Upgrade notes

  • No breaking changes. Drop-in upgrade for both standalone and embedded (@libredb/studio) usage.
  • The schema explorer now calls two new endpoints (/api/db/schema/list, /api/db/schema/relations). The legacy /api/db/schema endpoint is retained for full-schema consumers (diff, AI context).
  • Self-hosters on PostgreSQL with large schemas should see the table tree load in seconds instead of timing out.

Install

# Docker (GHCR — primary, no pull rate limits)
docker pull ghcr.io/libredb/libredb-studio:0.9.25

# npm package (embed in your app)
npm install @libredb/studio@0.9.25

# Helm
helm repo add libredb https://libredb.org/libredb-studio/
helm install libredb-studio libredb/libredb-studio

Thanks to everyone using and contributing to LibreDB Studio. Issues and PRs are welcome at https://github.com/libredb/libredb-studio.

Full Changelog: v0.9.19...v0.9.25

Assets 2
Loading