From: Tom Lane <tgl@sss.pgh.pa.us>
To: Alvaro Herrera <alvherre@2ndquadrant.com>
cc: Arthur Zakirov <zaartur@gmail.com>,
        Juan =?iso-8859-1?Q?Jos=E9_Santamar=EDa?= Flecha
 <juanjo.santamaria@gmail.com>,
        Tomas Vondra <tomas.vondra@2ndquadrant.com>,
        PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Subject: Re: Allow to_date() and to_timestamp() to accept localized names
In-reply-to: <20200123155552.GA20980@alvherre.pgsql>
References: <20200123155552.GA20980@alvherre.pgsql>
Comments: In-reply-to Alvaro Herrera <alvherre@2ndquadrant.com>
	message dated "Thu, 23 Jan 2020 12:55:52 -0300"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <11265.1579796123.1@sss.pgh.pa.us>
Date: Thu, 23 Jan 2020 11:15:23 -0500
Message-ID: <11266.1579796123@sss.pgh.pa.us>
Precedence: bulk

Alvaro Herrera <alvherre@2ndquadrant.com> writes:
> On 2020-Jan-22, Tom Lane wrote:
>> Arthur Zakirov <zaartur@gmail.com> writes:
>>> Shouldn't we just show all remaining string instead of truncating it? 

>> That would avoid a bunch of arbitrary decisions, for sure.
>> Anybody have an objection?

> I think it would be clearer to search for whitespace starting at the
> start of the bogus token and stop there.  It might not be perfect,
> particularly if any language has whitespace in a month etc name (I don't
> know any example but I guess it's not impossible for it to exist;
> Portuguese weekday names maybe?), but it seems better than either of the
> behaviors shown above.

I'm okay with that.  It won't work so well for cases like
"1-Januzry-1999", but it's still better than what happens now:

regression=# select to_date('1-Januzry-1999', 'DD MONTH YYYY');
ERROR:  invalid value "Januzry-1" for "MONTH"

That particular case could be improved by stopping at a dash ... but
since this code is also used to match strings like "A.M.", we can't
just exclude punctuation in general.  Breaking at whitespace seems
like a reasonable compromise.

			regards, tom lane