Feedback-ID: ia2694551:Fastmail
Date: Mon, 7 Aug 2023 11:10:44 +0200
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
To: Peter Smith <smithpb2250@gmail.com>
Cc: Dave Cramer <davecramer@gmail.com>, Tatsuo Ishii <ishii@sraoss.co.jp>,
	pgsql-hackers@lists.postgresql.org, nathandbossart@gmail.com
Subject: Re: Using defines for protocol characters
Message-ID: <20230807091044.jjgsl2rgheazaung@alvherre.pgsql>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <CAHut+PuE-3tc_6m06bwnjftk2KBM728ZSDhuL1_=vRCfEGMZ1A@mail.gmail.com>
Archived-At: <https://www.postgresql.org/message-id/20230807091044.jjgsl2rgheazaung%40alvherre.pgsql>
Precedence: bulk

On 2023-Aug-07, Peter Smith wrote:

> I guess, your patch would not be much different; you can still have
> all the nice names and assign the appropriate values to the enum
> values same as now, but using an enum you might also gain
> type-checking in the code and also get warnings for the "switch"
> statements if there are any cases accidentally omitted.

Hmm, I think omitting a 'default' clause (which is needed when you want
warnings for missing clauses) in a switch that handles protocol traffic
is not great, because the switch would misbehave when the network
counterpart sends a broken message.  I'm not sure we want to do that.
It could become a serious security problem if confronted with a
malicious libpq.

-- 
Álvaro Herrera         PostgreSQL Developer  —  https://www.EnterpriseDB.com/