On Fri, Dec 14, 2018 at 12:23:17AM -0500, Tom Lane wrote: > David Fetter <da...@fetter.org> writes: > > On Thu, Dec 13, 2018 at 07:43:52PM +0100, Peter Eisentraut wrote: > >> I'm not sure that this should be a run-time option. > > > Given the often onerous givens of installing new software on > > production machines, I'm thinking it actually should be. > > What's that have to do with it? This is a development option, > not a production option.
There are problems tab completion, e.g. its catalog queries, could cause in production that would be difficult to simulate elsewhere. > I don't think it should be in production builds at all, let alone be > invokable via a simple command-line switch. The notion that > ordinary users will want to use it is hogwash --- we'd be much more > likely to get complaints about psql spewing unexpected output. With utmost respect, by the time someone is turning on logging for psql, they're trying to diagnose a problem. If they're lucky, it's a problem that's happening during development of psql itself and gets caught during that process. I don't see "if they're lucky" is a helpful approach to observability. Best, David. -- David Fetter <david(at)fetter(dot)org> http://fetter.org/ Phone: +1 415 235 3778 Remember to vote! Consider donating to Postgres: http://www.postgresql.org/about/donate
