Full-text search

Problem

What becomes confusing, fragile, or impossible without understanding full-text search? This lesson answers that through explanation, a worked example, two runnable exercises, and a reference solution. No teacher-supplied worksheet is required.

WebAssembly brought a portable low-level compilation target to the browser; SQLite brought a mature relational engine into one embedded file.

Concept explanation

Full-text search belongs to “Persistent browser database”. full-text search is learned by comparing readable, constrained, and deliberately clever implementations against same evidence.

For full-text search, trace concrete input, state transition, output, and failure through an explicit low-level or relational boundary where memory, crossings, schemas, queries, and persistence costs can be measured.

JavaScript calls into a WASM module whose linear memory and host imports form an explicit boundary. SQLite executes queries over pages, indexes, and a planner. Apply that model to supplied normal, boundary, and failure cases; each case below names its input and expected evidence.

Example bank

Compare normal, boundary, failure, and cross-layer cases. Predict each observation before revealing the explanation.

Worked example

Start from supplied lab.sql. Focus: Solve full-text search three ways using identical fixture: Persist notes DB, reload, migrate schema, and export backup.

1. Run: sqlite3 :memory: < lab.sql
2. Save baseline evidence. Inspect module exports, memory, JS/WASM crossings, schemas, indexes, transactions, query plans, OPFS files, and recovery.
3. Boundary case: Run all three implementations against: Index multilingual full-text content and reopen in new session.
4. Failure case: Force every implementation through: Interrupt migration or import corrupt backup and recover previous version.

RESULT
All versions preserve Reload preserves data; migration is idempotent; backup round trip and failure recovery are demonstrated.; comparison records code size, predictability, and debugging cost. Starter-level baseline: Query plan reports use of observations_topic; final query prints baseline.

Starter material

Create lab.sql, paste this exact content, then run the command below.

CREATE TABLE observations (id INTEGER PRIMARY KEY, topic TEXT NOT NULL, result TEXT NOT NULL);
INSERT INTO observations(topic, result) VALUES ("full-text search", 'baseline');
CREATE INDEX observations_topic ON observations(topic);
EXPLAIN QUERY PLAN SELECT result FROM observations WHERE topic = "full-text search";
SELECT result FROM observations WHERE topic = "full-text search";

STOP / CLEANUPIn-memory SQLite exits after script. Stop any local web server with Ctrl+C.

Independent exercise

Compare three versions: full-text search

1. Create second case from blank file: Run all three implementations against: Index multilingual full-text content and reopen in new session.
2. Then create controlled failure: Force every implementation through: Interrupt migration or import corrupt backup and recover previous version.
3. Use WASM debugger · SQLite CLI · EXPLAIN QUERY PLAN · OPFS to prove behavior, then repair controlled failure.
4. Compare result against supplied acceptance checks and reference approach before marking complete.

Expected result

• All versions preserve Reload preserves data; migration is idempotent; backup round trip and failure recovery are demonstrated.; comparison records code size, predictability, and debugging cost.
• Query plan reports use of observations_topic; final query prints baseline.
• Controlled full-text search failure produces captured evidence; repair restores stated invariant without hiding error.

Hints

Solution

Attempt both exercises before opening reference approach.