Date: Fri, 8 Nov 2019 09:50:13 -0300
From: Alvaro Herrera <alvherre@2ndquadrant.com>
To: Mark Dilger <hornschnorter@gmail.com>
Cc: PostgreSQL-development <pgsql-hackers@postgresql.org>
Subject: Re: SPI refactoring
Message-ID: <20191108125013.GA26112@alvherre.pgsql>
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <cc57ea5c-592a-6c27-aca6-7ad2a166c379@gmail.com>
User-Agent: Mutt/1.10.1 (2018-07-13)
Precedence: bulk

On 2019-Nov-07, Mark Dilger wrote:

> From 113d42772be2c2abd71fd142cde9240522f143d7 Mon Sep 17 00:00:00 2001
> From: Mark Dilger <hornschnorter@gmail.com>
> Date: Thu, 7 Nov 2019 07:51:06 -0800
> Subject: [PATCH v1 1/5] Deprecating unused SPI error codes.
> 
> The SPI_ERROR_NOOUTFUNC and SPI_ERROR_CONNECT codes, defined in spi.h,
> were no longer used anywhere in the sources except nominally in spi.c
> where SPI_result_code_string(int code) was translating them to a cstring
> for use in error messages.  But since the system never returns these
> error codes, it seems unlikely anybody still needs to be able to convert
> them to a string.
> 
> Removing these from spi.c, from the docs, and from a code comment in
> contrib.  Leaving the definition in spi.h for backwards compatibility of
> third-party applications.

Because of PG_MODULE_MAGIC forcing a recompile of modules for each major
server version, there's no ABI-stability requirement for these  values.
If we were to leave the definitions in spi.h and remove the code that
handles them, then code could compile but at run-time it would produce
the "unrecognized" string.  Therefore I think it is better to remove the
definitions from spi.h, so that we can be sure that the code will never
be needed.

I didn't look at the other patches, but I suppose the same argument
applies to retaining their defines too.

-- 
Álvaro Herrera                https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services