From: Martin Duerst <duerst@...>
Date: 2007-11-08T16:46:21+09:00
Subject: Re: performance problem in 1.9

Yes. Ruby is open source, so you can check for yourself.
(one example is currently line 520 in encoding.c:

    if (rb_enc_mbmaxlen(enc) == rb_enc_mbminlen(enc)) {
        return (e - p) / rb_enc_mbminlen(enc);
    }

You can also run the test program from a previous mail
of yours with an US-ASCII encoding, and compare runnig
times with e.g. UTF-8 encoding. That will give a very
clear answer.

Regards,   Martin.

At 17:06 07/11/07, Wolfgang N〓dasi-Donner wrote:
>Martin Duerst schrieb:
>> The character counting problem you describe exists, but not for US-ASCII,
>> so this can't be the reason.
>I think it should be recognized, if an encoding has fixed size characters, so the character positions can be calculated -  but is this distinction implemented?
>
>Wolfgang Nテ。dasi-Donner
>


#-#-#  Martin J. Du"rst, Assoc. Professor, Aoyama Gakuin University
#-#-#  http://www.sw.it.aoyama.ac.jp       mailto:duerst@it.aoyama.ac.jp