> Additionally, I should think that JSON functions should not be included by default; they should be available as an option in the amalgamation and as a file that can be compiled separately if desired. (However, the -> and ->> syntax should be always included in the core SQLite, so that the extension will work.) You've just brought up a good argument *against* the new syntax, perhaps inadvertently: since they new syntax can be overridden with user-defined, functions, users moving from one sqlite3 app to another might have misconceptions about what those operators actually do. Much has been said about compatibility with PG and MySQL, and yet allowing the new syntax to be bound to UDFs means that the new syntax isn't even necessarily compatible within two different sqlite3 connections.