Message-ID: <gh-pgjdbc-pgjdbc-3724-c3052341841@github.com>
From: "vlsi (@vlsi)" <noreply+vlsi@github.com>
To: "pgjdbc/pgjdbc" <noreply+pgjdbc-pgjdbc@github.com>
Date: Wed, 09 Jul 2025 11:50:18 +0000
Subject: Re: [pgjdbc/pgjdbc] issue #3724: Regression: FATAL:  simple query "BEGIN" arrived before ending an extended query message
In-Reply-To: <gh-pgjdbc-pgjdbc-3724@github.com>
References: <gh-pgjdbc-pgjdbc-3724@github.com>
List-Id: <gh-pgjdbc-pgjdbc.github.com>
X-GitHub-Author-Login: vlsi
X-GitHub-Comment-Id: 3052341841
X-GitHub-Comment-Type: issue_comment
X-GitHub-Issue: 3724
X-GitHub-Repo: pgjdbc/pgjdbc
X-GitHub-Type: comment
X-GitHub-Url: https://github.com/pgjdbc/pgjdbc/issues/3724#issuecomment-3052341841
Content-Type: text/plain; charset=utf-8

Well, the validation was specific to pg-pool, and it is not clear if a missing sync message does any harm for the regular postgresql.

As per my understanding, it should not be that hard to inject the missing sync message at the driver level (it looks like a call to https://github.com/pgjdbc/pgjdbc/blob/master/pgjdbc/src/main/java/org/postgresql/core/v3/QueryExecutorImpl.java#L1569). However, it is not clear if we need to to extra work for transaction semantics.