Re: [PATCH] WIP Add ALWAYS DEFERRED option for constraints

From: Nico Williams <nico(at)cryptonector(dot)com>
To: Andreas Joseph Krogh <andreas(at)visena(dot)com>
Cc: pgsql-hackers(at)postgresql(dot)org
Subject: Re: [PATCH] WIP Add ALWAYS DEFERRED option for constraints
Date: 2017-10-03 20:22:37
Message-ID: 20171003202236.GP1251@localhost
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Tue, Oct 03, 2017 at 10:10:59PM +0200, Andreas Joseph Krogh wrote:
> +1.
>
> While we're in deferrable constraints land...; I even more often need
> deferrable conditional unique-indexes.
> In PG you now may have:
> ALTER TABLE email_folder ADD CONSTRAINT some_uk UNIQUE (owner_id, folder_type,
> name) DEFERRABLE INITIALLY DEFERRED;
>
> But this isn't supported:
> CREATE UNIQUE INDEX some_uk ON email_folder(owner_id, folder_type, name) WHERE
> parent_id IS NULL DEFERRABLE INITIALLY DEFERRED;
>
> Are there any plans to support this?

Not by me, but I can take a look and, if it is trivial, do it. At a
quick glance it does look like it should be easy enough to do it, at
least to get started on a patch.

If I can get some help with my current patch, I'll take a look :)

But yes, I'd like to have full consistency between CREATE and ALTER.
Everything that one can do with CREATE should be possible to do with
ALTER, including IF NOT EXISTS.

Nico
--

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Andrew Dunstan 2017-10-03 20:34:38 Re: datetime.h defines like PM conflict with external libraries
Previous Message Andreas Joseph Krogh 2017-10-03 20:10:59 Re: [PATCH] WIP Add ALWAYS DEFERRED option for constraints