Date: Thu, 19 Feb 2026 13:54:23 -0600
From: Nathan Bossart <nathandbossart@gmail.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: pgsql-hackers@postgresql.org
Subject: Re: assume availability of "inline" keyword
Message-ID: <aZdqb9df6oqsUq9m@nathan>
References: <aZdGbDaV4_yKCMc-@nathan>
 <2670103.1771528773@sss.pgh.pa.us>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <2670103.1771528773@sss.pgh.pa.us>
Archived-At: <https://www.postgresql.org/message-id/aZdqb9df6oqsUq9m%40nathan>
Precedence: bulk

On Thu, Feb 19, 2026 at 02:19:33PM -0500, Tom Lane wrote:
> All those usages are guarded by
> 
> #if defined(__GNUC__) || defined(__INTEL_COMPILER)
> 
> so we are only concerned with gcc-alikes, which reduces the risks
> of trouble.  However, I have a vague recollection that it is/was
> possible to put gcc in a mode where it only takes __inline__
> not inline.  This is likely something you get from -std=c90 or
> thereabouts, which would be irrelevant now.  But I'd suggest
> checking it's okay under -std=c11 (not gnu11), just to be sure.

Right.  I checked a couple of compilers and none seemed to mind this patch
with -std=c11.

-- 
nathan