Re: Generic Tagging: A Bold Proposal

From: Martin J. Duerst (mduerst@ifi.unizh.ch)
Date: Wed Jul 16 1997 - 08:44:05 EDT


On Tue, 15 Jul 1997, Pete Resnick wrote:

> Not all of us. Please go back and read the thread I posted in June with a
> subject of "CJK Tags - Fish or cut bait". I agree with you that full
> language tagging is not really necessary to put inside the text itself. The
> problem is that unlike any other plain text character set (ASCII, 8859-x,
> 2022), where you can always figure out at least some reasonable input
> method to use from the characters, Unicode does not allow you to do this
> for CJK. At least on the Macintosh, though I believe this is true on other
> systems that handle international text, the user at least has the option to
> have the system choose a default input method based on the character set
> being used, with no other external markup information. Likewise, I can
> easily tell from the Unicode script ranges which input method to use
> *except* for CJK.

Hello Pete,

Thanks for comming back to this. I agree that there are some
situations where your arguments may currently be applicable,
in particular on the Mac with Mac Scripts as currently implemented.

But I already have written a long answer to your mail more or less
showing that in general, CJK is nothing special. For example, if the
Mac had strictly separate scripts for Mac Roman (Western Europe) and
for Eastern Europe, instead of the current "sub-script" (or whatever
it actually is that makes Roman special in their implementation)
approach, or if they had used the same "sub-script" approach for
CJK, things would be quite different.

Also, I have not yet figured out how I could, with ResEdit, build
a keyboard that allows me to type in Latin letters from different
"codepages" (i.e. both Western Europe and Eastern Europe). That would
mean I would end up with pretty much the same problems for Latin.

All these things together seem to indicate that although your
problem of input method switching is an interesting case worth
discussing, it doesn't seem to provide very much evidence for
an absolute need for CJK (or other) lanugage tagging. Of course,
I'm glad to hear more from you about this.

Regards, Martin.



This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:35 EDT