Hi, On Wed, Nov 17, 2004 at 09:21:05AM -0500, Henry Spencer wrote:
> The decision clearly is arbitrary, with both approaches having advantages. > For example, in the context of "talk" etc. receiving only part of the > character temporarily, rendering an "a" without an accent is generally > more legible than rendering the accent without the "a". But if the combining accent character would precede the letter, then a talk client could just silently wait when it receives a combining prefix character, it wouldn't be necessary to print it. The difference is that in this case it knew that it would receive further bytes which will form one composed character. In the current way of unicode you can never know. > Do you say "a-acute" or "acute-a"? Of course I say "a-acute" and when I write (paper+pen) I first write an "a" and then put the accent on it. But IMHO how it is better to represent it with a byte stream is a different question. (Just as you said: > Data entry and data storage are two separate issues [...] .) > If combining accents are sent as they are typed, it would clearly be > useful to provide both, presumably specified by different keyboard > sequences. If the complete character is sent only after the keyboard > handler is satisfied (somehow) that it *is* complete, then backspace ought > to delete the whole thing. The problem is that when you press a backspace in talk, it's not the whole new line that is sent across the network, but an ascii 8 or 127, and the other party has to interpret it, which, of course, has no notion of keyboard handler, doesn't know how the sender typed those characters. Of course one may say it's a protocol-specific question not discussed by the Unicode standards... -- Egmont -- Linux-UTF8: i18n of Linux on all levels Archive: http://mail.nl.linux.org/linux-utf8/