Re: Mule autodetection is crap
>>>> In <87of7h811b.fsf(a)tleepslib.sk.tsukuba.ac.jp>
>>>> "Stephen J. Turnbull" <stephen(a)xemacs.org> wrote:
KY> The autodetection is one of the most important Mule
feature.
Yes, and it is _terribly_ broken in 21.4 from the point of view of
everyone in the world except Japanese and ISO-2022 multibyte users in
general. It's the attempt to make autodetection _safe_, let alone
useful, for the 95% of our users who aren't Japanese that is a major
motivation for rewriting Mule.
You are fair and right. Only Japanese people might be receiving
a benefit from the autodetection. However, it is a historical
inheritance of Mule version 1, Mule version 2, and possibly
Nemacs. I cannot bear the thought of being lost the feature.
We get patches to return things to Japan-centric normal from you
guys,
and they re-break things for the Europeans. That's not acceptable.
Suffering damages of people who aren't Japanese is not what I
want.
KY> There are other Mule-related problems as I reported in
the
KY> past.
You have the source, you have the data that uncovers the problem.
Feel free to fix it, just don't break things for other users.
The best I can personally offer you (I am not up to speed on 21.5
yet)
is that if you write a test for tests/automated/mule-tests.el, I'll
make sure it gets committed, and it will be an eternal irritant to Ben
(who hates test failures), until he fixes them.
That's a good idea. Emacs Lisp is one of the languages I can
speak well (hundreds times from English). Thanks for offering.
And I know Ben plans to get back to Mule soon after vacation, but
he's
got a couple more work spaces to commit first.
I heard it also before. I can believe all the other work of him
is extremely important for the future of XEmacs. But I wish he
don't leave Japanese XEmacs fan in the lurch.
--
Katsumi Yamaoka <yamaoka(a)jpl.org>