Amazing, especially - for me - that the FTS5 full-text search just works. Longer term, I am if it were possible to split the DB code into read and write parts and cross-compile only read part for delivery to the browser.
Lunr needs the full index on client though, right?
Being able to use fts-5 without the penalty of having to pull down the whole index make it work much better at larger scales, even with the penalty of additional network requests.
This could be genuinely interesting for tools like e.g. sphinx-doc, which currently has a client-side search that does indeed ship the entire index to the client.
There's probably a dataset size tradeoff with small enough number of documents, but even then this could have the option of if db is < x MB total, fetch all in async task and then use that.
Time to create the index for lunr limits the size of the data-set that can be used. If you have a lot of tiny documents, then it is more practical to scan the documents directly, rather than using the lunr index.
