Re: Improve error reporting in 027_stream_regress test

From: Michael Paquier <michael(at)paquier(dot)xyz>
To: Nazir Bilal Yavuz <byavuz81(at)gmail(dot)com>
Cc: Andres Freund <andres(at)anarazel(dot)de>, PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org>
Subject: Re: Improve error reporting in 027_stream_regress test
Date: 2025-07-19 06:05:56
Message-ID: aHs1xCvA47TOD4_8@paquier.xyz
Views: Whole Thread | Raw Message | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Fri, Jul 18, 2025 at 11:57:07AM +0300, Nazir Bilal Yavuz wrote:
> I added that as 0001. I used a shifting method for the 'tail'
> direction to not use too much memory. I found that there is
> 'File::ReadBackwards' in Perl but you need to install it, so I didn't
> use it.

Some old buildfarm may complain when it comes to that, like some of
Tom's stuff (?).

> 0001: 'Add trim_file() helper to Utils.pm' -> Which effectively does
> nothing, just adds a function to be used for a subsequent patch. This
> function accepts 'line_count' as an argument but
> 'PG_TEST_FILE_TRIM_LINES' environment variable overrides it. Should I
> document 'PG_TEST_FILE_TRIM_LINES' somewhere?

Documentation is required in regress.sgml, yes.

> 0002: 'Improve error reporting in 027_stream_regress test' -> Uses
> trim_file() function to improve error reporting by including the head
> and tail of regression.diffs directly in the failure message.

The structure is strange, it seems to me that we should target things
so as we have only one PG_TEST_FILE_TRIM_LINES defined in the tree,
not two with one local to 027.

> 0003: 'Check primary and standby are alive after regression tests in
> 027_stream_regress' -> Add test for checking status of primary and
> standby after the regression tests in 027_stream_regress. Also, it
> makes error reporting happen only if regression tests are failed and
> both primary and standby are alive.

This one looks acceptable to me, so applied to begin with something,
splitting things into two pieces for clarity with some tweaks. I have
changed things to use system_log() at the end, with fat-commas to link
the long options and their arguments.
--
Michael

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message vignesh C 2025-07-19 06:45:31 Re: Log prefix missing for subscriber log messages received from publisher
Previous Message Zhang Mingli 2025-07-19 05:15:04 Re: Even when the data is already ordered, MergeAppend still adds a Sort node