Re: dumping database privileges broken in 9.6

* Michael Paquier (michael(dot)paquier(at)gmail(dot)com) wrote:
> On Sat, Jul 16, 2016 at 4:46 AM, Stephen Frost <sfrost(at)snowman(dot)net> wrote:
> > Going through and doing testing now. Unfortunately, it doesn't look
> > like adding in testing of tablespaces into the TAP tests would be very
> > easy (the only TAP test that deals with tablespaces that I found was
> > pg_basebackup and that looks rather grotty..), so I'm not planning to do
> > that, at least not at this time.
>
> The cleanest way to handle that in PostgresNode would be to have a
> dedicated routine calling psql -c 'create tablespace' with tablespaces
> located in a folder $basedir/tbspace. And on top of that there should
> be the tablespace metadata saved in the context of the test, with a
> pair of (tbspc name, location). But I think that we'd need first
> stronger arguments (take more use cases) to introduce such an
> extension of the test APIs.

The way pg_basebackup handles this is to use TestLib::tempdir_short and
create a symlink with it, then to call psql to create the tablespace. I
don't have any problem using $basedir/tbspace instead though.

What I was thinking is that we'd add a 'create_tablespace' or similar
routine to PostgresNode and then make the pg_basebackup and pg_dump
tests use it. Otherwise, I suspect the next person who ends up writing
a 'create tablespace' into the TAP tests will use yet another location
and/or method.

Thanks!

Stephen