Horton

Re: BUG #17434: CREATE/DROP DATABASE can be executed in the same transaction with other commands

Tom Lane <tgl@sss.pgh.pa.us>

View thread View on postgr.es
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Yugo NAGATA <nagata@sraoss.co.jp>
Cc: pgsql-hackers@postgresql.org, "David G. Johnston" <david.g.johnston@gmail.com>, Bruce Momjian <bruce@momjian.us>
Date: 2022-11-09T16:17:29Z
Lists: pgsql-bugs, pgsql-hackers 
Yugo NAGATA <nagata@sraoss.co.jp> writes:
> Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> What do you think of the attached wording?

> It looks good to me. That describes the expected behaviour exactly.

Pushed that, then.

>> I don't think the pipeline angle is of concern to anyone who might be
>> reading these comments with the aim of understanding what guarantees
>> they have.  Perhaps there should be more about that in the user-facing
>> docs, though.

> I agree with that we don't need to mention pipelining in these comments,
> and that we need more in the documentation. I attached a doc patch to add
> a mention of commands that do internal commit to the pipelining section.
> Also, this adds a reference for the pipelining protocol to the libpq doc.

Hmm ... I don't really find either of these changes to be improvements.
The fact that, say, multi-table ANALYZE uses multiple transactions
seems to me to be a property of that statement, not of the protocol.

			regards, tom lane

Commits

  1. Rethink handling of [Prevent|Is]InTransactionBlock in pipeline mode.

  2. Doc: add comments about PreventInTransactionBlock/IsInTransactionBlock.

  3. Force immediate commit after CREATE DATABASE etc in extended protocol.