MIME-Version: 1.0
References: 
 <CAOYmi+nDZxJHaWj9_jRSyf8uMToCADAmOfJEggsKW-kY7aUwHA@mail.gmail.com>
 <CAOYmi+mRhhzGUvrcdickepAnsdaGbqhNcboNe4-YvgtkSzGNbQ@mail.gmail.com>
 <CA+hUKGLyS-GK_rkENaVbFVTa4VJ+reJxWtt+q4gmgNUXhptfYA@mail.gmail.com>
 <CAOYmi+k1q3feeMxfbaJA=+hx+XnOFQA0z2JU+0igA7fTUZTmoA@mail.gmail.com>
 <CA+hUKG+H1gwDh96jn5jB6Q3HyXrSC9x2y=uQJAthT8NLs6GN_Q@mail.gmail.com>
 <CAOYmi+n1xRNCDnwZzigXVk8V=+sr7ZzuGpJ0tAyozX-zQT19Gg@mail.gmail.com>
 <CAOYmi+mvk8Y7btYJhBzOGiNTY3cCpYZKjhA4-TP2Lkb=zOr4oQ@mail.gmail.com>
 <CA+hUKGKV5aM1K3gc_kAMtUq-BkD5AugLUCDVa-RMnbfFypALwQ@mail.gmail.com>
In-Reply-To: 
 <CA+hUKGKV5aM1K3gc_kAMtUq-BkD5AugLUCDVa-RMnbfFypALwQ@mail.gmail.com>
From: Thomas Munro <thomas.munro@gmail.com>
Date: Thu, 7 Aug 2025 14:43:17 +1200
Message-ID: 
 <CA+hUKGKt-5tB07T+ar+WaZCDuj-9CrnEUuy_O9giafAnaNzPwA@mail.gmail.com>
Subject: Re: [PATCH] OAuth: fix performance bug with stuck multiplexer events
To: Jacob Champion <jacob.champion@enterprisedb.com>
Cc: PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>,
 Daniel Gustafsson <daniel@yesql.se>,
	Peter Eisentraut <peter@eisentraut.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Archived-At: 
 <https://www.postgresql.org/message-id/CA%2BhUKGKt-5tB07T%2Bar%2BWaZCDuj-9CrnEUuy_O9giafAnaNzPwA%40mail.gmail.com>
Precedence: bulk

On Thu, Aug 7, 2025 at 1:45=E2=80=AFPM Thomas Munro <thomas.munro@gmail.com=
> wrote:
> I like the C TAP test.  PostgreSQL
> needs more of this.

I should add, I didn't look closely at that part since you said it's
not in scope for back-patching.  I'd like to, though, later.

I wonder if you would be interested in this attempt at centralised
infrastructure for unit testing our C code over here.  I'm not
suggesting it for your immediate problem, just noting the overlap:

https://www.postgresql.org/message-id/flat/CA%2BhUKG%2BajSQ_8eu2AogTncOnZ5m=
e2D-Cn66iN_-wZnRjLN%2Bicg%40mail.gmail.com

Basically I would like to be able to dump easy-to-write files into the
tree that say stuff like this =E2=86=93 and have the build scripts find the=
m,
build them and test them without all the module boilerplate stuff or a
running server (though that aspect is obviously not relevant for your
frontend case).  Like you find in googletest or various xunit-style
systems in other projects, but integrated with our TAP universe (or
whatever replaces it if we escape from Perl).  But I never got the
configure part of it working.

 PG_BEGIN_TESTS();
 ...
 PG_EXPECT_EQ(pg_preadv(fd, iov, 2, 11), 0);
 PG_EXPECT_EQ(pg_pread(fd, buffer, 10, 0), 10);
 PG_EXPECT_EQ_STR(buffer, "helloworld");
 ...
 PG_END_TESTS();