Horton
Menu

Re: pg17 issues with not-null contraints

Alvaro Herrera <alvherre@alvh.no-ip.org>

View thread
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
To: Justin Pryzby <pryzby@telsasoft.com>
Cc: Kyotaro Horiguchi <horikyota.ntt@gmail.com>, pgsql-hackers@lists.postgresql.org, Andrew Bille <andrewbille@gmail.com>
Date: 2024-04-17T17:45:13Z
Lists: pgsql-hackers

Attachments

On 2024-Apr-16, Justin Pryzby wrote:

> Yes, this fixes the issue I reported.

Excellent, thanks for confirming.

> BTW, that seems to be the same issue Andrew reported in January.
> https://www.postgresql.org/message-id/CAJnzarwkfRu76_yi3dqVF_WL-MpvT54zMwAxFwJceXdHB76bOA%40mail.gmail.com

That's really good news -- I was worried it would require much more
invasive changes.  I tested his case and noticed two additional issues,
first that we fail to acquire locks down the hierarchy, so recursing
down like ATPrepAddPrimaryKey does fails to pin down the children
properly; and second, that the constraint left behind by restoring the
dump preserves the "throaway" name.  I made pg_dump use a different name
when the table has a parent, just in case we end up not dropping the
constraint.

I'm going to push this early tomorrow.  CI run:
https://cirrus-ci.com/build/5754149453692928

-- 
Álvaro Herrera         PostgreSQL Developer  —  https://www.EnterpriseDB.com/

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Fix inconsistencies in error messages

  2. Fix restore of not-null constraints with inheritance

  3. Fix propagating attnotnull in multiple inheritance

  4. Doc: mention CREATE+ATTACH PARTITION with CREATE TABLE...PARTITION OF.

  5. Allow ATTACH PARTITION with only ShareUpdateExclusiveLock.