Content-Type: text/plain;
	charset=us-ascii
Mime-Version: 1.0 (Mac OS X Mail 16.0 \(3696.120.41.1.4\))
Subject: Re: doc issues in event-trigger-matrix.html
From: Daniel Gustafsson <daniel@yesql.se>
In-Reply-To: <Zfjm4nDEHMHlH8J6@paquier.xyz>
Date: Tue, 19 Mar 2024 10:34:32 +0100
Cc: jian he <jian.universality@gmail.com>,
 PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Content-Transfer-Encoding: quoted-printable
Message-Id: <19293801-0321-4B96-9F45-677F417BC516@yesql.se>
References: <CACJufxE_UAuxcM08BW5oVsg34v0cFWoEt8yBa5xSAoKLmL6LTQ@mail.gmail.com>
 <Zfjm4nDEHMHlH8J6@paquier.xyz>
To: Michael Paquier <michael@paquier.xyz>
Archived-At: <https://www.postgresql.org/message-id/19293801-0321-4B96-9F45-677F417BC516%40yesql.se>
Precedence: bulk

> On 19 Mar 2024, at 02:14, Michael Paquier <michael@paquier.xyz> wrote:
>=20
> On Tue, Mar 19, 2024 at 08:00:00AM +0800, jian he wrote:
>> I think the "X" and "-" mean in this matrix [0] is not very =
intuitive.
>> mainly because "X" tends to mean negative things in most cases.
>> we can write a sentence saying "X"  means this, "-" means that.
>>=20
>> or maybe Check mark [1] and Cross mark [2]  are more universal.
>> and we can use these marks.
>>=20
>> "Only for local objects"
>> is there any reference explaining "local objects"?
>> I think local object means objects that only affect one single =
database?

That's a bigger problem than the table representation, we never define =
what
"local object" mean anywhere in the EVT docs.  EV's are global for a =
database,
but not a cluster, so I assume what this means is that EVs for non-DDL =
commands
like COMMENT can only fire for a specific relation they are attached to =
and not
database wide?

> It is true that in Japan the cross mark refers to a negation, and
> that's the opposite in France: I would put a cross on a table in the
> case where something is supported.  I've never seen anybody complain
> about the format of these tables, FWIW, but if these were to be
> changed, the update should happen across the board for all the tables
> and not only one.

AFAICT we only have one other table with "X" denoting support, the =
"Conflicting
Lock Modes" table under Concurrency Control chapter, and there we simply =
leave
the "not supported" column empty instead of using a dash.  Maybe the =
simple fix
here is to make these tables consistent by removing the dash from the =
event
trigger firing matrix?

As a sidenote, the table should gain a sentence explaining why the login =
column
is missing to avoid confusion.

--
Daniel Gustafsson