Hacker News new | ask | show | jobs
by lovelearning 2437 days ago
Not an exhaustive list, but important differences off the top of my head:

- End-to-end search:

Algolia's offerings span both front-end (InstantSearch drop-in widgets) and back-end (actual search API). Simple applications can be built without ever talking to Algolia API at all because their widgets do it for you.

Atlas is all back-end - just a DB service with FTS on the side; left to you to integrate front-end.

- Configuration:

Algolia's dashboard GUI is where a lot of the configuration is done. Some configurations are not available at all via APIs. It's relatively simple.

Atlas requires more JSON-type configuration entries, and some knowledge of Lucene internals.

- Text analysis:

Algolia text tokenization pipeline is mostly a black-box but works fine most of the time. It exposes only a few settings like ascii-folding. It's fine for normal dictionary words, but has problems with domain-specific text (for example, people/place names, scientific terms, etc).

Atlas exposes many aspects of Lucene's analysis pipeline, but it does require knowledge of Lucene.

- Multilingual support:

Algolia supports all its features for ~70 languages.

Atlas analysis has to be configured separately for each language.

- Query syntax:

Algolia defaults to simple queries but the API supports a more complex query syntax with boolean operators and such.

Atlas has its own JSON query DSL that's related to Lucene's query syntax capabilities.

- Faceting:

Algolia faceting configuration and API are far simpler than Atlas's DSL.