Feedback-ID: ie0a040ee:Fastmail
Message-ID: <cbf88173-fce9-49f1-a9ee-29f8e982b98a@eisentraut.org>
Date: Fri, 17 Oct 2025 17:23:22 +0200
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: Change initdb default to the builtin collation provider
To: Jeff Davis <pgsql@j-davis.com>, pgsql-hackers@postgresql.org
References: <e4ac16908dad3eddd3ed73c4862591375a3f0539.camel@j-davis.com>
Content-Language: en-US
From: Peter Eisentraut <peter@eisentraut.org>
In-Reply-To: <e4ac16908dad3eddd3ed73c4862591375a3f0539.camel@j-davis.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Archived-At: 
 <https://www.postgresql.org/message-id/cbf88173-fce9-49f1-a9ee-29f8e982b98a%40eisentraut.org>
Precedence: bulk

On 11.10.25 02:48, Jeff Davis wrote:
> The builtin provider uses code point order, i.e. memcmp(), so the
> final result display order is less human-friendly. For instance, 'Z'
> comes before 'a'.
> 
> That problem is annoying, but*much* easier to fix than the other
> factors. The user might add a COLLATE clause to the final ORDER BY, or
> perform the sort in the application layer or presentation layer.

I remain violently opposed to this idea.  I don't understand how it 
could be acceptable to just not provide a good display order by default 
and have everyone rewrite their queries.

> ICU is better than libc in a lot of ways:
> 
> * Better performance
> * Platform-independent
> * Easier to manage it as a separate library
> 
> But fundamentally, I don't think it's a great default, because it
> favors final result display order at the risk of primary key
> inconsistencies.

I don't understand.  We have a versioning system for ICU collations? 
Does it not work?