Date: Wed, 29 Dec 2021 10:30:15 -0600
From: Justin Pryzby <pryzby@telsasoft.com>
To: Peter Smith <smithpb2250@gmail.com>
Cc: pgsql-hackers@lists.postgresql.org
Subject: Re: PublicationActions - use bit flags.
Message-ID: <20211229163014.GO24477@telsasoft.com>
References: 
 <CAHut+Pt2hjRJxfuF9_GUcBVg=nhk3G+WNsNkPtP2d+4CF4e4Pg@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: 
 <CAHut+Pt2hjRJxfuF9_GUcBVg=nhk3G+WNsNkPtP2d+4CF4e4Pg@mail.gmail.com>
User-Agent: Mutt/1.9.4 (2018-02-28)
Archived-At: 
 <https://www.postgresql.org/message-id/20211229163014.GO24477%40telsasoft.com>
Precedence: bulk

On Mon, Dec 20, 2021 at 11:18:41AM +1100, Peter Smith wrote:
> For some reason the current HEAD PublicationActions is a struct of
> boolean representing combinations of the 4 different "publication
> actions".
> 
> I felt it is more natural to implement boolean flag combinations using
> a bitmask instead of a struct of bools. IMO using the bitmask also
> simplifies assignment and checking of said flags.

Peter E made a suggestion to use a similar struct with bools last year for
REINDEX.
https://www.postgresql.org/message-id/7ec67c56-2377-cd05-51a0-691104404abe@enterprisedb.com

Alvaro pointed out that the integer flags are better for ABI compatibility - it
would allow adding a new flag in backbranches, if that were needed for a
bugfix.

But that's not very compelling, since the bools have existed in v10...
Also, the booleans directly correspond with the catalog.

+       if (pubform->pubinsert) pub->pubactions |= PUBACTION_INSERT;

This is usually written like:

	pub->pubactions |= (pubform->pubinsert ? PUBACTION_INSERT : 0)

-- 
Justin