Message-ID: <75c43ae49dba00a3af7a4d0cb06498bef9508c4c.camel@cybertec.at>
Subject: Re: Change initdb default to the builtin collation provider
From: Laurenz Albe <laurenz.albe@cybertec.at>
To: Robert Haas <robertmhaas@gmail.com>, Jeff Davis <pgsql@j-davis.com>
Cc: pgsql-hackers@postgresql.org
Date: Thu, 12 Mar 2026 21:32:43 +0100
In-Reply-To: 
 <CA+Tgmob7WJ3iETZea=P_eQ2-74q6xE_Vqnz074M5w5VD=dKURA@mail.gmail.com>
References: <e4ac16908dad3eddd3ed73c4862591375a3f0539.camel@j-davis.com>
	 <47e1b4f72fe732c5ae85c6cf2c085b4e99a10120.camel@j-davis.com>
	 <CA+TgmobnOn9ipEUcthLmxvv0ZKBWyCcc048qqgqy1X+tszw_Cg@mail.gmail.com>
	 <4309879ac305b1cf6b4d7b5fb85bc7b62c6ab768.camel@j-davis.com>
	 <CA+TgmoYPfiz6_QjkiKykwziFDZFzPVf_x77rChYBu=OGVOmbYw@mail.gmail.com>
	 <be03b4f8e5ed147a3e733d9a73ddeb51e6b83ad4.camel@j-davis.com>
	 <CA+Tgmob7WJ3iETZea=P_eQ2-74q6xE_Vqnz074M5w5VD=dKURA@mail.gmail.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.58.3 (3.58.3-1.fc43) 
MIME-Version: 1.0
Archived-At: 
 <https://www.postgresql.org/message-id/75c43ae49dba00a3af7a4d0cb06498bef9508c4c.camel%40cybertec.at>
Precedence: bulk

On Thu, 2026-03-12 at 10:04 -0400, Robert Haas wrote:
> So maybe the people who are saying that defaulting to C is fine and
> that making people make an explicit choice if they want something else
> are right. but my personal guess is that we will make a bunch of
> people unhappy.

Defaulting to C will make a bunch of people unhappy, I agree.
However, that is a kind of problem that they will discover early on
during development or testing, and it is easily remedied for those
columns where alphabetical sorting order matters.

But a good number of people are already unhappy because they have
index corruption (today I dealt with another support case).
And the people who are diligent enough to know that they should
reindex after an OS update are unhappy because of the additional
down time.

I maintain that the second kind of unhappiness weighs heavier,
and I find this need to reindex to be one of the most embarrassing
flaws in an otherwise great piece of software.

This is a value judgement, and I may be wrong.

> If even a small fraction of users create a
> database using "C" unintentionally and load a terabyte of data into it
> before realizing that all their text indexes are sorting "wrong", I
> suspect that's not going to be much fun.

Hm.  So reindexing a large table after an OS upgrade is acceptable,
but reindexing a large table after a bulk load is not?  The problem
you describe could have been avoided with a bit of testing, but the
need to reindex after an OS upgrade sometimes cannot be avoided.

> Obviously, I could be wildly incorrect. Maybe people will just be
> super-happy about faster sorting and life will be great.

I have not experienced sort speed as a frequent problem, and the
life with the C collation won't be all that great.  But I believe
that it would be better than a life with index corruption.

Yours,
Laurenz Albe