Date: Mon, 14 Nov 2016 12:32:59 -0800
From: Andres Freund <andres@anarazel.de>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Amit Kapila <amit.kapila16@gmail.com>,
	Ashutosh Sharma <ashu.coek88@gmail.com>,
	Robert Haas <robertmhaas@gmail.com>,
	"pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>
Subject: Re: Re: [COMMITTERS] pgsql: Change the way that LWLocks
 for extensions are allocated.
Message-ID: <20161114203259.fw3b3cmb67k2zl4z@alap3.anarazel.de>
References: <20160830023140.tsesenk7roajx3ak@alap3.anarazel.de>
 <CAA4eK1JOjkdXYtHxh=2aDK4VgDtN-LNGKY_YqX0N=YEvuzQVWg@mail.gmail.com>
 <CAA4eK1Jt5fYe2jnQ7XXDrpy3Z4W80UYhQa9SvCyXz+7mcoN8Gw@mail.gmail.com>
 <CAE9k0P=tew0UXETJfM5WQjqNS0k7xYZEFWMZCY8ZuEqRSD1uUA@mail.gmail.com>
 <CAA4eK1LJRtV=jBa46F6+=s7cf-x7ajFAraYBdAxaCjZ0m_FQ0w@mail.gmail.com>
 <20161112130408.kfsxcbkqo2b3nvuq@alap3.anarazel.de>
 <22432.1478968242@sss.pgh.pa.us>
 <20161112163508.5uqlaozeyr3l56ae@alap3.anarazel.de>
 <22899.1478968932@sss.pgh.pa.us>
 <20161114201410.ctgglamtdu2u5yj3@alap3.anarazel.de>
MIME-Version: 1.0
Content-Type: multipart/mixed; boundary="ae2y3mjzi3vicwx7"
Content-Disposition: inline
In-Reply-To: <20161114201410.ctgglamtdu2u5yj3@alap3.anarazel.de>
Precedence: bulk
Sender: pgsql-hackers-owner@postgresql.org


--ae2y3mjzi3vicwx7
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline

On 2016-11-14 12:14:10 -0800, Andres Freund wrote:
> On 2016-11-12 11:42:12 -0500, Tom Lane wrote:
> > Andres Freund <andres@anarazel.de> writes:
> > > On 2016-11-12 11:30:42 -0500, Tom Lane wrote:
> > >> which is a rather blatant waste of cycles. I would suggest an explicit
> > >> do-nothing installcheck rule rather than the hack you came up with here.
> >
> > > I had that at first, but that generates a warning about overwriting the
> > > makefile target - which afaics cannot be fixed.
> >
> > Hm.  What about inventing an additional macro NO_INSTALLCHECK that
> > prevents pgxs.mk from generating an installcheck rule?  It's not
> > like we don't have similar issues elsewhere, eg contrib/sepgsql.
>
> Well, that one seems a bit different.  Seems easy enough to do. Do we
> want to make that macro that prevents installcheck from being defined,
> or one that forces it to be empty? The former has the disadvantage that
> one has to be careful to define a target, to avoid breaking recursion
> from the upper levels.

Oh, that disadvantage doesn't actually exist, because installcheck is a
.PHONY target...

I've for now not added a variant that prevents plain 'make check' from
doing anything. I guess it could be useful when a contrib module wants
to run tests via something else than pg_regress?

Patch attached.

Andres

--ae2y3mjzi3vicwx7
Content-Type: text/x-patch; charset=us-ascii
Content-Disposition: attachment;
 filename="0001-Provide-NO_INSTALLCHECK-for-pgxs.patch"