On 2020-01-24 12:44, Alvaro Herrera wrote:
> On 2020-Jan-24, Juan José Santamaría Flecha wrote:
>
>> There is an open patch that will make the normalization functionality user
>> visible [1]. So, if a user can call to_date(normalize('01 ŞUB 2010'), 'DD
>> TMMON YYYY') I would vote to drop the normalization logic inside this patch
>> altogether.
>
> I was reading the SQL standard on this point, and it says this (4.2.8
> Universal character sets):
>
> An SQL-implementation may assume that all UCS strings are normalized
> in one of [Unicode normalization forms].
>
> which seems to agree with what you're saying.
When you interact with glibc locale functions, you essentially have to
assume that everything is in NFC. When using ICU locale functions
(which we don't here, but just for the sake of argument), then I would
expect that ICU does appropriate normalization itself when I call
icu_what_month_is_this(string) returns int. So I think it is
appropriate to not deal with normalization in this patch.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
