search

Filters are better when the URL reflects state. I use a small Stimulus controller that updates the query string as filters change, then triggers a Turbo visit (often with data-turbo-action='replace'). This gives shareable URLs and correct back-button

For search-as-you-type, I keep the server in charge and use a small Stimulus controller to debounce form submission. The controller listens to input events, waits ~250ms, then triggers a normal Turbo form submit. The server responds with index.turbo_s

Full reindexes can be long and fragile. Add checkpoints (last processed id), process in batches, and make it resumable. That turns a scary operation into a routine one.

Full-text search finds documents matching text queries. PostgreSQL tsvector stores processed documents optimized for search. I use tsquery for search queries with operators—AND, OR, NOT. GIN indexes on tsvector columns enable fast search. Text search

Elasticsearch provides powerful full-text search capabilities. Spring Data Elasticsearch offers repository abstraction similar to JPA. @Document annotates entity classes with index mapping. @Field customizes field types and analyzers. Queries use meth

Client-side debounce is best done in Stimulus (not in view helpers). This controller submits the nearest form after a short pause, while letting Turbo handle the navigation and frame replacement.

Real-time search enhances discoverability but naive implementations hammer the server with requests. I use Turbo Frames to scope search results and a Stimulus controller to debounce input events, only sending requests after typing pauses. The search f

Haystack provides unified search across backends (Elasticsearch, Solr, Whoosh). I define search indexes mapping models to searchable fields. It handles full-text search, faceting, and highlighting. The SearchQuerySet API is similar to Django's ORM. I

For Postgres search beyond trivial ILIKE, maintain a tsvector column and a GIN index. Update it via trigger or application logic. This keeps search fast and predictable even as your dataset grows.

Search inputs that fire API requests on every keystroke create poor UX and waste server resources. Debouncing delays the search until typing pauses, reducing requests dramatically. I combine a controlled input component with a custom useDebounce hook