On Tue, Aug 16, 2022 at 10:10:55AM -0400, Tom Lane wrote: > There is nothing principled about assuming that the first word > after SELECT is a function name. It'd be even less principled to > provide tab completion only for function names beginning with > "pg_". So this idea seems like a wart rather than something > anybody would think is a nice improvement. While I understand that there is nothing that would suggest it, is there any reason why providing a thing that can legally be there would be bad idea? I understand that someone might want to enter pg_or_not_pg (column name from some table), or perhaps pg.some_column_name - but what is the harm of providing pg_* functions for pg_<tab>? Specifically, I'd ask what is the harm of increasing what tab completion can do by a lot - for example, make it tab-complete fields from all tables. And make it possible to tab-complete column name anywhere in where clause. But function name in select would be (for me(!)) great first step, and I can't really see the drawbacks, aside from using developer time to work on it. Best regards, depesz
