[throwmea]

I can really agree on the first point but know too little about the second one. It is indeed a problem but i think in most cases keywords or tags should be enough to implement. Im not sure but even full text search should not take that long for plaintext/md notes.

[PaulHoule]

It's not a problem in terms of speed or CPU consumption, it's a problem in terms of relevance. It's when the relevant document you are looking for is result #34 or when the relevant document doesn't appear at all.

People will say "just use Lucene, bro" but often Lucene's search ranking is awful. (I used to laugh at "just use Solr, bro" because Solr's search scales about as well as Bitcoin when you add more shards.)

As for tags that depends on people putting accurate tags on. People hate to hear it, but tags need to have an ontology behind them to be really effective. For instance there is an anime imageboard that has tags like "purple hair", "green eyes" and these are close to useless for an image that has two people in it.

[throwmea]

fair point i agree with that thanks for explaining