On Wednesday, March 18, 2015, WadeDare4703 <wadedare4...@comcast.net> wrote:
> I don't understand. What is wrong with having a schema which holds no > data? Schemas are cheap. > > ------------------------------ > *From: *"Tim Uckun" <timuc...@gmail.com > <javascript:_e(%7B%7D,'cvml','timuc...@gmail.com');>> > *To: *"pgsql-general" <pgsql-general@postgresql.org > <javascript:_e(%7B%7D,'cvml','pgsql-general@postgresql.org');>> > *Sent: *Wednesday, March 18, 2015 6:56:55 PM > *Subject: *[GENERAL] Name spacing functions and stored procedures > > What do you guys do to namespace your functions so that they are not > jumbled in with the system functions and also somewhat hierarchically > organized. > > Obviously it's possible to create schemas for different namespaces but > that can lead to a lot of schemas which hold no data. The other way is to > simply name your functions like _lib_etl_csv_import_weird_data_format but > that's not too much fun either. > > Just curious how other people deal with the issue. > > Create an extension? Barring that a dedicated schema without user data has desirable properties whether you manage it manually or via an extension. The former ways allows you to write a migration script using drop schema and install everything from scratch. While not useable in all cases it can be done. I do wish schemas had some additional properties, like being able to readily specify (if not enforce) private versus public functions, for visibility during schema browsing, or creating a hierarchy which too would be visible in browsing. This is what I think you are wishing for as well. Something like Java packages. I think you can do the naming but tooling support is non-existent and, if it's not readily provided in core and used by psql or pgadmin it's unlikely other tools would want to write their own. David J.
