Re: Extensions not dumped when --schema is used

On Tue, Apr 13, 2021 at 08:00:34AM -0700, Noah Misch wrote:
> On Tue, Apr 13, 2021 at 02:43:11PM +0900, Michael Paquier wrote:
>>> - If extschema='public', "pg_dump -e plpgsql --schema=public" includes
>>>   commands to dump the relation data.  This surprised me.  (The
>>>   --schema=public argument causes selectDumpableNamespace() to set
>>>   nsinfo->dobj.dump=DUMP_COMPONENT_ALL instead of DUMP_COMPONENT_ACL.)
>>
>> This one would be expected to me.  Per the discussion of upthread, we
>> want --schema and --extension to be two separate and exclusive
>> switches.  So, once the caller specifies --schema we should dump the
>> contents of the schema, even if its extension is not listed with
>> --extension.
>
> I may disagree with this later, but I'm setting it aside for the moment.
>
> This isn't a problem of selecting schemas for inclusion in the dump.  This is
> a problem of associating extensions with their pg_extension_config_dump()
> relations and omitting those extension-member relations when "-e" causes
> omission of the extension.

At code level, the decision to dump the data of any extension's
dumpable table is done in processExtensionTables().  I have to admit
that your feeling here keeps the code simpler than what I have been
thinking if we apply an extra filtering based on the list of
extensions in this code path.  So I can see the value in your argument
to not dump at all the data of an extension's dumpable table as long
as its extension is not listed, and this, even if its schema is
explicitly listed.

So I got down to make the behavior more consistent with the patch
attached.  This passes your case.  It is worth noting that if a
table is part of a schema created by an extension, but that the table
is not dependent on the extension, we would still dump its data if
using --schema with this table's schema while the extension is not
part of the list from --extension.  In the attached, that's just the
extra test with without_extension_implicit_schema.

(By the way, good catch with the duplicated --no-sync in the new
tests.)

What do you think?
--
Michael