On Mon, 11 Jan 1999 22:39:59 +0200 (EET), Jarkko Hietaniemi
<jhi(_at_)iki(_dot_)fi> said:
You don't explain, why.
jhi> How about getting the initial data much more easier? Now when I start
jhi> to look for non-Western-European data I will find it in encodings X,
jhi> Y, and Z, and I can almost bet a small sum that none of those won't be
jhi> Unicode.
Sure. And you can be happy when you don't get them as GIFs :-)
But this brings me to the idea, that for later debugging you should
indeed leave the sources around. Sometimes more than one source. But
then, the first and canonical step should be Unicodization to enter
a common ground.
jhi> I am just keeping the data structure
jhi> open enough to allow for multiple different encodings. For the
jhi> moment, because it's easier and more useful both for me and for users,
jhi> I encode in Latin-X. Later, when Unicode has conquered the world,
jhi> it's no problem to add UTF-8 encoding.
It's not a matter if Unicode will conquer the world. It's the matter,
that Unicode solves a problem that you plan to treat with a
complicated data structure that appears superfluous to me. If you can
jhi> This is complicated?
jhi> my %weekday =
jhi> (
jhi> 'af' =>
jhi> {
jhi> 'ISO 8859-1' =>
jhi> [
jhi> 'Sondag',
jhi> 'Maandag',
jhi> 'Dinsdag',
jhi> 'Woensdag',
jhi> 'Donderdag',
jhi> 'Vrydag',
jhi> 'Saterdag',
jhi> ],
jhi> },
It is complicated if you fill the hash with 300 encodings. And then
you need the logic to supply ISO 8859-2 although you have it only as
ISO 8859-1. And then you start two-way converting. The middle encoding
most likely being -- you guess it -- something Unicodisch. Been there,
done that.
jhi> Andreas, you are wearing me out: soon I'll give in to your UTF-8 preaching
:-)
I hope so. I've already wasted a lot of time by doing it wrong myself. :-/
--
andreas
