Re: Fucking Mulål

"Stephen J. Turnbull" <[email protected]> writes:

> I see you're not running latin-unity yet.

I'm not.  I don't like the current level of Unicode support in Mule,
so I'm staying away from such stuff while I can.  But I've been known
to change my mind about such matters, so don't take this as a generic
"NO" to ever trying it out.

> What happened is Mule managed to wedge itself into the wrong coding
> system somehow.  If you tell Mule to save, it does so.  If the
> selected coding system can't handle a character, it replaces it with
> a tilde, silently.

I know that -- but what I don't understand is *how* -- in high-level
terms -- I managed to get that to happen.  From a high-level (usage)
perspective, it's a bizarre bug, one that causes data corruption.  And
this was a *text* file, not a theoretic binary someone, sometime would
want to edit.

I lost half an hour to restore the file, and I can only thank the
relevant deities that the Croatian language has only five letters that
fall outside the ASCII range.

> latin-unity implements (for all ISO Latin sets supported by XEmacs)
> a safety test, it allows you to configure charsets you always find
> acceptable so it won't bug you if it knows you're happy, it undoes
> the Mule bogosity of thinking that Latin-1 NO-BREAK SPACE != Latin-2
> NO-BREAK SPACE, it supports the Euro, and best of all, even though
> we know it has no bugs ;-), if you should manage to exercise one
> anyway, you know where to find the programmer and make him pay....

It sounds very cool, yes.

> It also may be slow if you've got large numbers of non-ASCII
> characters.  I've got some ideas for further optimization, so let me
> now if efficiency is a problem.

I don't think it's a problem for Croatian.  Cyrillic-using languages
might be affected, though.