Re: Ought to use heap_multi_insert() for pg_attribute/depend insertions?

From: Daniel Gustafsson <daniel(at)yesql(dot)se>
To: Thomas Munro <thomas(dot)munro(at)gmail(dot)com>
Cc: Andres Freund <andres(at)anarazel(dot)de>, pgsql-hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Ought to use heap_multi_insert() for pg_attribute/depend insertions?
Date: 2019-07-09 11:07:15
Views: Raw Message | Whole Thread | Download mbox | Resend email
Lists: pgsql-hackers

> On 8 Jul 2019, at 00:02, Thomas Munro <thomas(dot)munro(at)gmail(dot)com> wrote:
> On Wed, Jun 12, 2019 at 1:21 AM Daniel Gustafsson <daniel(at)yesql(dot)se> wrote:
>> The patch is still rough around the edges (TODO’s left to mark some areas), but
>> I prefer to get some feedback early rather than working too far in potentially
>> the wrong direction, so parking this in the CF for now.
> Hi Daniel,
> Given the above disclaimers the following may be entirely expected,
> but just in case you weren't aware:
> t/ fails with this patch applied.

I hadn’t seen since I had fat-fingered and accidentally run the full tests in a
tree without assertions. The culprit here seems to an assertion in the logical
decoding code which doesn’t account for heap_multi_insert into catalog
relations (which there are none now, this patch introduce them and thus trip
the assertion). As the issue is somewhat unrelated, I’ve opened a separate
thread with a small patch:

The attached v3 also has that fix in order to see if the cfbot is happier with

cheers ./daniel

Attachment Content-Type Size
catalog_multi_insert-v3.patch application/octet-stream 45.0 KB

In response to


Browse pgsql-hackers by date

  From Date Subject
Next Message Peter Eisentraut 2019-07-09 11:09:38 Re: Add parallelism and glibc dependent only options to reindexdb
Previous Message Thomas Munro 2019-07-09 11:03:18 Re: Introduce timeout capability for ConditionVariableSleep