On Mon, Jan 30, 2012 at 11:12:09AM -0500, Tom Lane wrote:
> Yes, that's the intended behavior. The purpose of the
> configuration-table feature is to dump data that is needed for an
> extension to operate properly --- for instance postgis has some
> auxiliary tables that are best treated as part of the schema. If you
> think the contents of a table are not effectively schema information,
> then you shouldn't mark it as pg_extension_config_dump.
but the it is *NOT POSSIBLE* do dump data of the table with pg_dump. AT
ALL. regardless of options.
> > I have since made some tests, and it looks like the dumping thing is
> > fixed in 9.2devel from git HEAD, which seems to suggest that it will
> > work sanely in 9.1.3 - so the point is kind of moot.
> What tests were those exactly? I'm not aware of any agreed changes in
> this area.
1. create table with extension
2. try to dump it's content with pg_dump
3. try to dump schema of database
4. try to dump schema of another table (not from extension)
5. drop extension, load another extension, this time, with marking table
with pg_extension_config_dump
repeat steps 2,3,4.
the way pg works in git HEAD is much more sensible.
Best regards,
depesz
--
The best thing about modern society is how easy it is to avoid contact with it.
http://depesz.com/