Refining the Single Binary: Schema Migration, Cursor Pagination, and Zero-Config HTTPS

Over the past month, I focused on making Moul resilient in production rather than piling on new features. A single-binary backend only works if schema changes, security certificates, and binary updates don't require external scripts or complex operational machinery.

In traditional cloud stacks, small operational tasks rely on external tooling. You write bash scripts, configure Certbot daemons, set up CI deployment runners, or execute manual migration scripts over SSH. In a single-binary architecture, these primitives must live inside the binary itself. It should run on a cheap virtual server without accumulating maintenance debt.

Here is how the latest set of changes makes that vision more reliable.

Self-Updating Binaries and Simplified Setup

Installing and maintaining a compiled binary should be as straightforward as running it.

• One-Command Installation: Added install.sh to download and place both moul (the TUI console) and moul-dev (the server engine) directly into your system path.
• In-Place Self-Updates: Introduced moul update and moul-dev update. The engine checks GitHub releases, downloads compiled binaries, verifies checksums, and replaces the running executable¹.

Deployments no longer require building from source on tiny target servers or setting up custom package repositories.

Dynamic Schema Synchronization

Moul allows defining database schemas—called Mouls—dynamically via REST endpoints or the TUI. But schemas evolve over time, and altering SQLite tables safely requires care.

• Non-Destructive Column Sync: The engine now inspects SQLite PRAGMA table_info (SyncMoulTableColumns) on startup and schema modification, executing non-destructive ALTER TABLE statements when fields are added or modified.
• Strict Relation Enforcement: Added validation rules for foreign key relations and introduced Select/Enum field types to keep data clean at the database boundary.
• Timestamp Standardization: Every collection now automatically manages created_at and updated_at fields using standardized RFC3339 UTC timestamps.

It works without manual SQL migrations. The database adapts to your data definitions safely.

Data Flow: Server-Side Filtering and Cursor Pagination

Offset-based pagination degrades as tables grow larger. Skimming past thousands of rows wastes database IO and slows down client applications.

• Cursor-Based Pagination: Replaced offset pagination in ListRecords with cursor pagination using indexed primary keys. Query response times remain flat regardless of table depth.
• Server-Side Filtering and Sorting: Added server-side evaluation for queries, allowing clients to slice and order datasets before records hit the network.
• Lifecycle Webhooks: Added webhook dispatches for create, update, and delete operations. External services can react to database changes in real time over simple HTTP POST payloads.

Infrastructure Primitives: TLS, Telemetry, and Flags

To run independently on a bare server, the binary must manage its own network perimeter and health tracking.

• Zero-Config TLS: Integrated CertMagic for automatic ACME TLS certificate acquisition and renewal. Point a domain to your server IP, and moul-dev serves HTTPS out of the box.
• System Telemetry: Integrated native metrics collection directly into the server, exposing CPU, memory, and disk telemetry without running third-party monitoring daemons.
• Feature Flags: Added a light feature flag store managed directly from the moul TUI, allowing you to toggle application behavior on the fly.

Craftsmanship in software isn't about accumulating features. It's about removing friction and building systems that endure. Moul continues to get smaller, faster, and more self-reliant.