Re: Another --> Re: GtkMovementStep of GtkTextView

[Steve Underwood <steveu coppice org>]

Mark Leisher wrote:
> 
>     Owen> You might want to try it out and see if it does what you want.  The
>     Owen> one thing that may not be desirable for some languages is that if
>     Owen> you are sitting after a compound grapheme and hit delete, it deletes
>     Owen> the entire grapheme.
> 
>     Owen> It may be desirable to make this configurable.. while I can imagine
>     Owen> that if you have a complex cluster in an Indic language you may just
>     Owen> want to undo the last typed key if you make a typo, Europeans will
>     Owen> no doubt not enjoy the experience of having ü<Delete> produce u.
> 
> Having dabbled with issues like this for some time, we found that in most
> cases, deleting the whole cluster was acceptable and often expected.  But
> Korean and Vietnamese users, though they could live with more general
> behavior, wanted the last key typed to be deleted and not the whole group of
> keys that made up the Hangul/vowel respectively.

The locally produced Indic word processors I saw working in India had a
hybrid behaviour. If you hit the backspace key at any time while typing
in some text, only the last entered keystroke was removed. However, in
any other context the whole grapheme was removed. I was told this was
very intentional behaviour, so that no screwy conbination of elements
could ever appear in a character cell. If the construction of a grapheme
is an IME issue, then only whole graphemes would exist in the committed
text, and there should be no objection to zapping them whole.

I can understand Koreans wanting to backspace element by element, but
doing that in Vietnamese seems odd - its just like Owen's example of a
backspace simply stripping the umlaut from a U.

Regards,
Steve