From: Tom Lane <tgl@sss.pgh.pa.us>
To: Heikki Linnakangas <hlinnakangas@vmware.com>
cc: "Boguk, Maksym" <maksymb@fast.au.fujitsu.com>,
 pgsql-hackers@postgresql.org
Subject: Re: UTF8 national character data type support WIP patch and list of
 open issues.
In-reply-to: <522594E8.2050106@vmware.com>
References: <A756FAD7EDC2E24F8CAB7E2F3B5375E918E2D5B8@FALEX03.au.fjanz.com>
 <522594E8.2050106@vmware.com>
Comments: In-reply-to Heikki Linnakangas <hlinnakangas@vmware.com>
 message dated "Tue, 03 Sep 2013 10:51:04 +0300"
Date: Tue, 03 Sep 2013 09:58:19 -0400
Message-ID: <15548.1378216699@sss.pgh.pa.us>
Precedence: bulk
Sender: pgsql-hackers-owner@postgresql.org

Heikki Linnakangas <hlinnakangas@vmware.com> writes:
> On 03.09.2013 05:28, Boguk, Maksym wrote:
>> Target usage:  ability to store UTF8 national characters in some
>> selected fields inside a single-byte encoded database.

> I think we should take a completely different approach to this. Two 
> alternatives spring to mind:

> 1. Implement a new encoding.  The new encoding would be some variant of 
> UTF-8 that encodes languages like Russian more efficiently.

+1.  I'm not sure that SCSU satisfies the requirement (which I read as
that Russian text should be pretty much 1 byte/character).  But surely
we could devise a variant that does.  For instance, it could look like
koi8r (or any other single-byte encoding of your choice) with one byte
value, say 255, reserved as a prefix.  255 means that a UTF8 character
follows.  The main complication here is that you don't want to allow more
than one way to represent a character --- else you break text hashing,
for instance.  So you'd have to take care that you never emit the 255+UTF8
representation for a character that can be represented in the single-byte
encoding.  In particular, you'd never encode ASCII that way, and thus this
would satisfy the all-multibyte-chars-must-have-all-high-bits-set rule.

Ideally we could make a variant like this for each supported single-byte
encoding, and thus you could optimize a database for "mostly but not
entirely LATIN1 text", etc.

			regards, tom lane


-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers