Thread: More thoughts on drop tablespace

The same thing may also apply to the default tablespace of a database as 
well...

Chris

Christopher Kings-Lynne wrote:

> The same thing may also apply to the default tablespace of a database 
> as well...
>
Well at least the database can be queried easily for usage of that 
tablespace.
For the namespace issue, it would help if *some* object would be created 
with that namespace (e.g. an empty directory or file with the namespace 
oid), so that pg_tablespace_databases() could detect it.

Regards,
Andreas

> Well at least the database can be queried easily for usage of that 
> tablespace.

Yes, that's the easy part to fix.  You'd just set the dattablespace back 
to 0 when you dropped the tablespace.

> For the namespace issue, it would help if *some* object would be created 
> with that namespace (e.g. an empty directory or file with the namespace 
> oid), so that pg_tablespace_databases() could detect it.

Seems like the only way.

Chris

Christopher Kings-Lynne <chriskl@familyhealth.com.au> writes:
> The same thing may also apply to the default tablespace of a database as 
> well...

No, because it will always contain some files (at the very least, the
db's pg_class *must* live there).  See prior thread about exactly
this issue.
        regards, tom lane