Re: SPI isolation changes

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: Seino Yuki <seinoyu(at)oss(dot)nttdata(dot)com>
Cc: Heikki Linnakangas <hlinnaka(at)iki(dot)fi>, pgsql-hackers(at)lists(dot)postgresql(dot)org
Subject: Re: SPI isolation changes
Date: 2023-06-30 15:06:15
Message-ID: 1758726.1688137575@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Seino Yuki <seinoyu(at)oss(dot)nttdata(dot)com> writes:
> I also thought that using SPI_start_transaction would be more readable
> than using SPI_commit/SPI_rollback to implicitly start a transaction.
> What do you think?

I think you're trying to get us to undo commit 2e517818f, which
is not going to happen. See the threads that led up to that:

Discussion: https://postgr.es/m/3375ffd8-d71c-2565-e348-a597d6e739e3@enterprisedb.com
Discussion: https://postgr.es/m/17416-ed8fe5d7213d6c25@postgresql.org

It looks to me like you can just change the transaction property
settings immediately after SPI_start_transaction if you want to.
Compare this bit in SnapBuildExportSnapshot:

StartTransactionCommand();

/* There doesn't seem to a nice API to set these */
XactIsoLevel = XACT_REPEATABLE_READ;
XactReadOnly = true;

Also look at the implementation of SPI_commit_and_chain,
particularly RestoreTransactionCharacteristics.

regards, tom lane

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Heikki Linnakangas 2023-06-30 15:07:16 On /*----- comments
Previous Message Tom Lane 2023-06-30 15:00:09 Re: Assert !bms_overlap(joinrel->relids, required_outer)