On Tuesday, October 11, 2016, Josh Berkus <j...@agliodbs.com> wrote: > On 10/10/2016 10:50 PM, Dave Page wrote: > > > > > > On Monday, October 10, 2016, Josh Berkus <j...@agliodbs.com > <javascript:;> > > <mailto:j...@agliodbs.com <javascript:;>>> wrote: > > > One thing which would improve this is to have setup.py take an > optional > > --directory argument. > > > > > > How so? It gets the path from the config, the same way the app does. If > > you separate them, it's bound to go wrong for some users. > > Ah, so I can drop a path into config_local? >
Yes, per the example I sent yesterday. config_local.py (and config_distro.py) are sourced by config.py. config_distro is read first, and can be used by packagers to override distro-specific options, and then config_local is read to allow users to do further customisation. > > In that case, configuring that and running setup.py is what needs to > happen for the RPM ... > > Not really - the problem there is that the same package needs to support both desktop and server modes - and that's proving not to work so well at present. The security side of the problem should be solved by the patch I sent earlier which ensures that all installations will have unique keys. There will remain conflicts between modes though, with regard to some of the paths. I suspect what we'll have to end up doing is have everything work out of the box for desktop mode, but have users tweak config_local.py for server mode. -- Dave Page Blog: http://pgsnake.blogspot.com Twitter: @pgsnake EnterpriseDB UK: http://www.enterprisedb.com The Enterprise PostgreSQL Company