Re: Should we get rid of custom_variable_classes altogether?

David Fetter <david@fetter.org> writes:
> Perhaps it's best to document this usage and include the warning for
> those less "bright," as you term them.   I'd be less tempted to call
> them "not bright" and more tempted to think they might assume
> PostgreSQL already takes care of cleaning this up, but whatever.

Who's that dim?  D'oh.

Another compromise might be to allow for defining variable in any class
from the configuration files but restrict that to existing classes from
the SET command.  Wait, that's exactly what happens as soon as there's
no explicit custom_variable_classes, right?

So we're talking about people with configuration file editing and reload
powers, not about anyone who can connect.  I think that's ok.

Regards,
-- 
Dimitri Fontaine
http://2ndQuadrant.fr     PostgreSQL : Expertise, Formation et Support