From erik@sran8.sra.co.jp Fri Sep  6 05:18:19 1991
Received: from mcsun.EU.net by dkuug.dk via EUnet with SMTP (5.64+/8+bit/IDA-1.2.8)
	id AA20676; Fri, 6 Sep 91 05:18:19 +0200
Received: from srawgw.sra.co.jp by mcsun.EU.net with SMTP;
	id AA22226 (5.65a/CWI-2.106); Fri, 6 Sep 1991 05:18:36 +0200
Received: from srava.sra.co.jp by srawgw.sra.co.jp (5.64WH/1.4)
	id AA16825; Fri, 6 Sep 91 12:17:53 +0859
Received: from sran8.sra.co.jp by srava.sra.co.jp (5.64c/6.4J.6-BJW)
	id AA04769; Fri, 6 Sep 91 12:17:57 +0900
Received: from localhost by sran8.sra.co.jp (5.65/6.4J.6-SJ)
	id AA09588; Fri, 6 Sep 91 12:18:20 +0900
Return-Path: <erik@sran8.sra.co.jp>
Message-Id: <9109060318.AA09588@sran8.sra.co.jp>
Reply-To: erik@sra.co.jp
From: erik@sra.co.jp (Erik M. van der Poel)
To: wg15rin@dkuug.dk
Subject: substitute in LC_COLLATE
Date: Fri, 06 Sep 91 12:18:02 +0900
Sender: erik@sran8.sra.co.jp
X-Charset: ASCII
X-Char-Esc: 29

> > Delete lines 2180-2183.		(substitute)
> 
> Does that mean that the substitute command is eliminated?
> Is that OK with the japanese?

As the unfortunate owner of the action item to check whether we need
the `substitute' functionality, I guess I should respond to this,
although I am by no means an "official" representative of "the
Japanese". (I encourage the Japanese to comment.)

As I said before, the size of the LC_COLLATE section of the source fed
to localedef can be reduced by using the substitute command for
certain Japanese characters. If collating elements are used instead of
substitutes, this section will be larger. However, it is not clear to
me that this size is really important. It may be more important for
the runtime collation to be efficient.

I am not currently able to gauge the performance impact, since I don't
have an implementation. Hopefully, my action item does not require me
to actually implement the collation! :-) If someone could test this on
their implementation, I would be grateful. (I still have copies of the
two LC_COLLATE tables, one that uses substitute and one that uses
collating elements.)

On the other hand, I'm also quite willing to accept a version of
POSIX.2 that does not include `substitute'. If some implementor
somewhere in the world really feels the need for substitute, he/she
can implement it, and put the specs in a national profile, or perhaps
in the next version of POSIX.2.


Best Regards,
EvdP