On 3/19/2014 9:17 PM, J. Leslie Turriff wrote:
Perhaps it might be useful to be able to distinguish between an "editing
mode" and a "composition mode": editing mode would be active when a document
is first loaded into the editor, when the editor has no keystroke history to
consult, and in this mode the backspace key would merely remove text "glyph
by glyph", so to speak, as happens with ASCII text; composition mode would
be active when keystrokes have been recorded in a buffer, so that backspace
could be used to "unstroke" the original strokes; the "unstroke" operations
would mimic the order in which the originals were entered, even if the editor
had optomized the composition.
It's more complicated than that.
Many editors don't (always) support "micro" undo. At some point,
keystrokes (or their result) are coalesced and an undo will delete
entire words or phrases, perhaps entire bullet items on a slide.
If done right, this will feel natural. If I've made edits to my document
in three places, inserting the same word, then it feels natural to
"undo" these as whole words (and not slavishly by keystroke - including
all the false starts and backspace keys).
At the current caret position, one would expect the undo to be less
aggressive and act more like a backspace. But in that case the user
would (roughly) remember the keystrokes that just happened, so inverting
the sequence feels more natural.
That same memory is why backspacing by composition step (keystroke) is
appealing - you intuitively know how many wrong keys were pressed. But
many user interfaces do not support that. Composing SMS with the T9
interface will let you erase characters from the composed string, but
will not revert to earlier word-guesses, so you can't cycle back, except
by erasing until the beginning and then starting over. For that
interface, the upside is that sometimes breaking a word apart by
freezing the composition of the leading part, erasing parts that don't
fit and then composing the remainder is the most efficient way to get
around some limitations of the composition method.
Whatever the details, the design of an ideal user interface should not
drive, or worse, dictate the character encoding - nor should the reverse
be true.
A./
_______________________________________________
Unicode mailing list
Unicode@unicode.org
http://unicode.org/mailman/listinfo/unicode
