Re: BackgroundPsql swallowing errors on windows

Daniel Gustafsson <daniel@yesql.se>

From: Daniel Gustafsson <daniel@yesql.se>

To: Andres Freund <andres@anarazel.de>

Cc: PostgreSQL Hackers <pgsql-hackers@postgresql.org>, Noah Misch <noah@leadboat.com>

Date: 2025-02-14T12:35:40Z

Lists: pgsql-hackers

Commits

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

tests: BackgroundPsql: Fix potential for lost errors on windows
- bb2bf22761e1 13.21 landed
- c21ebcca7965 14.18 landed
- 70b650d18553 15.13 landed
- 4c9397c374d9 16.9 landed
- f61769adaa62 17.5 landed
- 8b886a4e3488 18.0 landed
Improve handling of empty query results in BackgroundPsql::query()
- 70291a3c66ec 18.0 cited
Extend Cluster.pm's background_psql() to be able to start asynchronously
- ba08edb06545 18.0 cited

> On 13 Feb 2025, at 18:39, Andres Freund <andres@anarazel.de> wrote:

> The banner being the same between queries made it hard to understand if a
> banner that appeared in the output was from the current query or a past
> query. Therefore I added a counter to it.

+	my $banner = "background_psql: QUERY_SEPARATOR $query_cnt";
+	my $banner_match = qr/(^|\n)$banner\r?\n/;
+	$self->{stdin} .= "$query\n;\n\\echo $banner\n\\warn $banner\n";
+	pump_until(
+		$self->{run}, $self->{timeout},
+		\$self->{stdout}, qr/$banner_match/);

Won't this allow "QUERY_SEPARATOR 11" to match against "QUERY_SEPARATOR 1"?
It's probably only of academic interest but appending an end-of-banner
character like "_" or something after the query counter should fix that.

> For debugging I added a "note" that shows stdout/stderr after executing the
> query, I think it may be worth keeping that, but I'm not sure.

I think it could be worth it, +1 for keeping it until it's beeb proven a
problem somewhere.

> This was a rather painful exercise.

I have no trouble believing that.

--
Daniel Gustafsson