Thor Lancelot Simon <tls@rek.tjls.com> writes:

> On Mon, Sep 06, 2004 at 11:27:11AM +0900, Curt Sampson wrote:
> > 
> > I am not sure I clearly understand Soda-san's reasons for wanting to
> > ensure that "ja_JP.Shift_JIS" or "ja_JP.sjis" will not work, and those
> > parts that I may understand, I disagree with. So I will let him make his
> > argument for going with route #1 instead of route #2 (if that is indeed
> > what he is trying to argue.)
> 
> Encouraging the proliferation of nonstandard names for objects is almost
> _always_ a bad idea.  All it does is send you down a slippery slope on which
> you must add hack after hack for further compatibility with even more
> busted user input in the future, as others try to be more "helpful" and
> "flexible" than you are.

I personally don't see why it is such a bad thing to accomodate user
input, as long as the results are clearly unambiguous.  We can be
strict with computer input, but user input is a different matter.  But
that is just my opinion.

In any case, let me ask a different question.  Right now there is
support for aliases in src/share/locale/ctype/Makefile, and that
support is used to make zh_TW.BIG5 an alias for zh_TW.Big5.  Should
that aliasing support be removed?  Which one of zh_TW.BIG5 and
zh_TW.Big5 should be retained?

Ian