Feedback-ID: ie0a040ee:Fastmail
Message-ID: <9d4fb5e5-b1c5-4e56-865d-93ba6a390022@eisentraut.org>
Date: Mon, 26 Jan 2026 11:20:04 +0100
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: get rid of Pointer type, mostly
To: David Geier <geidav.pg@gmail.com>, Chao Li <li.evan.chao@gmail.com>
Cc: Robert Haas <robertmhaas@gmail.com>, Tom Lane <tgl@sss.pgh.pa.us>,
 pgsql-hackers <pgsql-hackers@postgresql.org>
References: <4154950a-47ae-4223-bd01-1235cc50e933@eisentraut.org>
 <955002.1764000569@sss.pgh.pa.us>
 <CA+Tgmobo9ZMC0h1Z1Te9whf7q3LN7wUkzFLyuxfMZ4dyTNqKnw@mail.gmail.com>
 <957744.1764002037@sss.pgh.pa.us>
 <CA+TgmobhKe5WJybZ0EGQ85Qkxr-KLx4uW2_hFqt0iPbYq-af7Q@mail.gmail.com>
 <970261.1764008763@sss.pgh.pa.us>
 <CA+Tgmob==vi7MXZC9EhTGZ2YiuWkQuAGZm=H6M=9zPN2tp4s3A@mail.gmail.com>
 <987824.1764013105@sss.pgh.pa.us>
 <CA+TgmobFfk2y9jXPzzme-3qQCJC9LFnd2G9RBMCE9haVJkCQ+w@mail.gmail.com>
 <7c7e1e59-284b-49f6-a496-17dd4170f979@gmail.com>
 <2b929da1-dbb5-476f-a878-899b1e9e80f2@gmail.com>
 <8B037F29-163E-4446-BA94-7553C9ECA0F1@gmail.com>
 <c6047830-fc89-4a25-8e7a-c50a94ad6e0d@eisentraut.org>
 <0d5ca69a-91c7-4efb-a53a-9948a4f32717@gmail.com>
Content-Language: en-US
From: Peter Eisentraut <peter@eisentraut.org>
In-Reply-To: <0d5ca69a-91c7-4efb-a53a-9948a4f32717@gmail.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Archived-At: <https://www.postgresql.org/message-id/9d4fb5e5-b1c5-4e56-865d-93ba6a390022%40eisentraut.org>
Precedence: bulk

On 08.12.25 12:27, David Geier wrote:
> On 08.12.2025 12:14, Peter Eisentraut wrote:
>> I was planning to proceed with Dagfinn's patch set.  Here is what is
>> currently remaining of the patch series.  I haven't fully processed
>> everyone's comments in this thread, so they might not be reflected in
>> these patches.
> Dagfinn's patch set turns Pointer into void *. I thought we had agreed
> to use something like GinExtraPointer in the GIN code (see my patch set).
> 
> Either way is fine by me. It just seemed like the majority of commenters
> where in favor of keeping some type in the GIN code for readability.

After further consideration, I think it is best to leave this alone as 
is.  One particular problem is that Pointer is part of the documented 
API of GIN, and if we either remove it or rename it, then this would 
create a bunch of work for extensions to keep their code compiling 
across major versions.