Re: [COMMITTERS] pgsql: Add COMMENT and SECURITY LABEL support for publications and subs

From: Stephen Frost <sfrost(at)snowman(dot)net>
To: Peter Eisentraut <peter(dot)eisentraut(at)2ndquadrant(dot)com>
Cc: pgsql-hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: [COMMITTERS] pgsql: Add COMMENT and SECURITY LABEL support for publications and subs
Date: 2017-04-04 13:59:01
Message-ID: 20170404135901.GD9812@tamriel.snowman.net
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-committers pgsql-hackers

Peter,

* Peter Eisentraut (peter(dot)eisentraut(at)2ndquadrant(dot)com) wrote:
> On 3/25/17 12:24, Stephen Frost wrote:
> >> Add COMMENT and SECURITY LABEL support for publications and subscriptions
> >
> > Isn't this missing psql tab completion, and pg_dump support? And
> > regression tests for the latter?
>
> I have added the tab completion support.

Thanks!

> Attached is a patch that adds the pg_dump support, but I'm struggling to
> make the tests work. Could you take a look? Problem one I'm seeing is
> that the tests assert that there are no comments in the post-data
> section, which is no longer the case here.

If that's the case (and is intended), then you'll need to remove the
'section_post_data' entry from the COMMENTS catch-all 'unlike' and move
that into the 'unlike' for each of the COMMENT tests which were
depending on the catch-all to handle that.

The other approach is to create a new catch-all which catches COMMENTs
that are not pub/sub and have the other COMMENT tests use that but have
a different catch-all for the pub/sub entries or just have all of the
tests covered by them. There are a few examples of this approach
already (off-hand, I think there's one related to blobs).

> Problem two is that
> subscriptions are not dumped by default, so those new tests fail, but I
> don't understand why the existing tests about subscriptions work so far.

There are (or were?) some tests which explicitly use
'--include-subscription'. You can see how each test's pg_dump is run in
the hash at the top.

I thought we were changing that anyway though? To dump subscriptions by
default but to have them dumped in a 'not enabled' fashion? In fact, I
thought that had already happened, but I might be thinking of something
else.

Thanks!

Stephen

In response to

Responses

Browse pgsql-committers by date

  From Date Subject
Next Message Peter Eisentraut 2017-04-04 16:51:36 pgsql: Adjust min/max values when changing sequence type
Previous Message Peter Eisentraut 2017-04-04 13:28:34 Re: [COMMITTERS] pgsql: Add COMMENT and SECURITY LABEL support for publications and subs

Browse pgsql-hackers by date

  From Date Subject
Next Message Noah Misch 2017-04-04 14:07:17 v10 Open Item Ownership
Previous Message David Steele 2017-04-04 13:53:14 Re: Re: PATCH: pageinspect / add page_checksum and bt_page_items(bytea)