Message-ID: <504bc10a9ad8d143f505e66be2a0060e50b45677.camel@j-davis.com>
Subject: Re: Order changes in PG16 since ICU introduction
From: Jeff Davis <pgsql@j-davis.com>
To: Alexander Lakhin <exclusion@gmail.com>, Tom Lane <tgl@sss.pgh.pa.us>
Cc: Daniel Verite <daniel@manitou-mail.org>, Andrew Gierth
 <andrew@tao11.riddles.org.uk>, Peter Eisentraut
 <peter.eisentraut@enterprisedb.com>, Sandro Santilli <strk@kbt.io>, Regina
 Obe <lr@pcorp.us>, pgsql-hackers@lists.postgresql.org
Date: Tue, 16 May 2023 09:47:51 -0700
In-Reply-To: <de1e72d9-0a89-6d96-1995-db44f5476956@gmail.com>
References: <543e4ce6-8205-4aea-b54b-12b9ef747bf6@manitou-mail.org>
	 <654a49f7ff7461bcf47be4181430678d45f93858.camel@j-davis.com>
	 <37520ec1ae9591f83132f82dbd625f3fc2d69c16.camel@j-davis.com>
	 <483826.1683582475@sss.pgh.pa.us>
	 <aa6cac622417eda631a8a53287905fda1a963723.camel@j-davis.com>
	 <2098874d-c111-41e4-9063-30bcf135226b@gmail.com>
	 <91c0cdd2a402d59f8d6eaac40862d6f68e0b937c.camel@j-davis.com>
	 <de1e72d9-0a89-6d96-1995-db44f5476956@gmail.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.44.4-0ubuntu1 
MIME-Version: 1.0
Archived-At: <https://www.postgresql.org/message-id/504bc10a9ad8d143f505e66be2a0060e50b45677.camel%40j-davis.com>
Precedence: bulk

On Tue, 2023-05-16 at 19:00 +0300, Alexander Lakhin wrote:
> I'm not sure about the proposed change in icu_from_uchar(). It seems
> that
> len_result + 1 bytes should always be enough for the result string
> terminated
> with NUL. If that's not true (we want to protect from some ICU bug
> here),
> then the change should be backpatched?

I believe it's enough and I'm not aware of any bug there so no backport
is required.

I added checks in places that were (a) checking for U_FAILURE; and (b)
expecting the result to be NUL-terminated. That's mostly callers of
uloc_getLanguage(), where I was not quite paranoid enough.

There were a couple other places though, and I went ahead and added
checks there out of paranoia, too. One was ucnv_fromUChars(), and the
other was uloc_canonicalize().

Regards,
	Jeff Davis