From: Tom Lane <tgl@sss.pgh.pa.us>
To: Noah Misch <noah@leadboat.com>
cc: David Rowley <dgrowleyml@gmail.com>,
        PostgreSQL Developers <pgsql-hackers@lists.postgresql.org>
Subject: Re: Recent failures on buildfarm member hornet
In-reply-to: <2397925.1602107299@sss.pgh.pa.us>
References: <1830491.1601944935@sss.pgh.pa.us>
 <20201007014507.GB1422897@rfd.leadboat.com>
 <2040113.1602035809@sss.pgh.pa.us>
 <20201007021941.GD1422897@rfd.leadboat.com>
 <2233910.1602090798@sss.pgh.pa.us>
 <CAApHDvovu2O+SLerPK2fCqKKP-q6NWinzACvJkMQ9zZNEiCKJg@mail.gmail.com>
 <20201007212056.GB1498955@rfd.leadboat.com>
 <2397925.1602107299@sss.pgh.pa.us>
Comments: In-reply-to Tom Lane <tgl@sss.pgh.pa.us>
	message dated "Wed, 07 Oct 2020 17:48:19 -0400"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <2398542.1602108196.1@sss.pgh.pa.us>
Date: Wed, 07 Oct 2020 18:03:16 -0400
Message-ID: <2398543.1602108196@sss.pgh.pa.us>
Precedence: bulk

I wrote:
> I'm tempted to propose the attached small code rearrangement, which
> might dissuade the compiler from thinking it can get away with this.
> While I concur with your point that an old xlc version might not be
> that exciting, there could be other compilers doing the same thing
> in the future.

After thinking about it a bit more, I'm not even convinced that what
xlc seems to be doing is illegal per C spec.  There are no sequence
points within

    return list_make2(list_concat(directargs, orderedargs),
                      makeInteger(ndirectargs));

and therefore there's an argument to be made that the compiler
doesn't have to care whether any side-effects of list_concat() occur
before or after the evaluation of makeInteger(ndirectargs).
If the potential side-effects of list_concat() can be disregarded
until the end of this statement, then the code change is perfectly legal.

Maybe some very careful language-lawyering could prove differently,
but it's not as open-and-shut as one could wish.  So now I'm thinking
that we need this patch anyway, xlc or not.

			regards, tom lane