html_url,issue_url,id,node_id,user,created_at,updated_at,author_association,body,reactions,issue,performed_via_github_app https://github.com/simonw/datasette/issues/607#issuecomment-548055544,https://api.github.com/repos/simonw/datasette/issues/607,548055544,MDEyOklzc3VlQ29tbWVudDU0ODA1NTU0NA==,9599,2019-10-30T18:37:44Z,2019-10-30T18:37:52Z,OWNER,".Hi @zeluspudding You're running your search queries using the ""contains"" filter, which uses a `like` query under the hood. SQL `like` queries are generally slow because they force a full table scan. You can add an index on the column but it will only speed up prefix queries, like `... where name like 'apple%'` - they won't help if you are searching for text further along the string. Instead, you should take a look at SQLite's FTS - full text indexing feature. You can build a FTS index against a column and dramatically speed up searches for words within that column. This documentation should help get you started: https://datasette.readthedocs.io/en/stable/full_text_search.html","{""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",512996469,