= lnx 0.9, the fast search engines like Elasticsearch & AAlgolia alternative is out! + Tech Talk =

👋 Hello all, it's been a little while. I'm glad to finally be able to say that 0.9 of lnx is finally out after a few months of inactivity, I got slightly de-railed building the website backend for a friend, and doing a rather enjoyable tech talk about lnx in June

For context, lnx is a fast alternative to Elasticsearch & AAloglia written in Rust and built upon the amazing Tantivy library. Providing a good mix of utility and ease of use without sacrificing on performance.In fact, the fast-fuzzy system makes fuzzy text searching almost as cheap as regular text search, seeing potentially tens of thousands of queries per second on a 10-core machine

The code is located at httpsgithub.com/lnx-search/lnx with various other tools and libraries included under the organisation, and docs available at httpsdocs.lnx.rs
The full changelog is available @ httpsgithub.com/lnx-search/lnx/releases/tag/0.9.0A small blog post about it is also available @ httpschillfish8.ghost.io/whats-new-in-lnx-0-9/
I also did a tech talk at a Rust London meetup earlier this month (went massively over the time limit as well) which may be of interest to anyone bored out of their minds: httpswww.youtube.com/watch?v=kzCYbZjJcTk
Congrats!
Comparisons with your closest search neighbors would be highly informative, Toshi being the most important one since it’s also based on Tantivy. Maybe include Sonic as well, since it’s an alternative to ElasticSearch

I’d recommend against comparing lnx to Algolia, since that’s a different kind of speciality in search, and Tantivy isn’t headed in that direction. MeiliSearch has that particular use case very well covered already

I agree comparing against Sonic would be interesting, I'm slightly less interested in comparing Toshi, simply because its development is pretty much stalled at this point, but would still be interesting nonetheless

I’d recommend against comparing lnx to Algolia, since that’s a different
kind of speciality in search, and Tantivy isn’t headed in that
direction. MeiliSearch has that particular use case very well covered
already

Although I can see your reasoning behind it, I do disagree with a few points, I love Meilisearch, we use it for prototyping some search systems and it's great that it's easy to set up and relevant. However, our biggest limiter with though is it's only really suitable for small-scale setups and million-ish documents or less due to its almost exponential indexing time (We generally expect indexing 100 million docs to take around an hour or less, rather than a week or more 😅)

So although we're not directly competing with Meilisearch in the small index size area, we are competing with it in searching millions of documents or greater, although Algolia might not be able to cope with either, so it may be a redundant comparison!
As for the quote about tantivy, I do have to strongly disagree with that, there isn't anything particularly tying Tantivy to any particular area of search, it's essentially Lucene in Rust and provides all of the tooling you need to need to make a user-facing search like Meilisearch and the likes (Especially with the latest versions which add efficient term aggregations), a good example of this Etsy which uses Tantivy

I think arguably elastic search tries to do too much, and full-text-search is still its core offering (for the ElasticSearch product at least), and in a benchmark