Accelerators missing from dialogs in KDE 4.3

[Richard Dale]

On Friday 31 July 2009 08:45:11 pm Ingo Klöcker wrote:
> On Wednesday 29 July 2009, Chusslove Illich wrote:
> > > [: Richard Dale :]
> > > I personally think the whole concept is broken of having letters
> > > underlined in menu items to denote which key you need to press with
> > > alt. And the underlines look really ugly [...] Is there really no
> > > better way of invoking menu items from the keyboard than by using
> > > alt + 'a fairly random character' ?
> >
> > I agree with you, though not only from aesthetic point of view, but
> > also due to difficulty of resolving clashes when programming, and yet
> > more so when translating, as dicussed so far in the thread.
> >
> > I personally contend that the "mnemonic" bit about accelerators is
> > overrated, even going as far as labeling it useless. I've no hard
> > proof to back this up; I think the proof (or disproof) could be
> > provided by measuring and comparing amount of accelerator usage among
> > German and Japanese users (both countries at similar development
> > level, German users have all-mnemonic accelerators, Japanese
> > all-nonmnemonic).
> >
> > Asuming for a moment that mnemonics in this case is indeed useless,
> > my perfect handling of accelerators would look like this:
> >
> > * Programmers manually assign accelerators only for some widely used
> > labels (e.g. "Case Sensitive"). All other accelerators are
> > automatically assigned. (Same as Stephan argues.)
> >
> > * However, manually assigned accelerators are not specified by
> > putting anything into the text label itself, but in some other way.
> >
> > * Translators thus do not see nor deal with accelerators at all;
> > accelerator markers fall out of realm of translation (just like the
> > shortcuts are now).
> >
> > * Users see accelerators only if they want to use them, and not shown
> > all the time. Possibly something happens when the user presses the
> > Alt key in applicable context?
> >
> > Middle of the road solution for the few manually assigned
> > accelerators would be to still specify accelerator in the label
> > (either for implementation reasons, or to allow translators to adapt
> > them), but in a highly formulaic fashion, such that it's stripped
> > before the label is actually shown. Perhaps the CJK system "...(&X)"
> > would do, though it looks too much like part of the text (which it no
> > longer would be!); perhaps "...~&X", or whatever.
>
> I'm wondering whether a completely different approach would be possible:
> Substring matching, i.e. after opening a menu one types part of the
> menu entry one wants to execute. The matching letters of all matching
> menu entries would be highlighted (as with spotlight on Mac). Tab would
> allow switching between several matching entries. Space and Enter would
> execute the selected option. This approach should also work for Asian
> languages.
>
> Of course, this approach does not provide the efficiency of single-key
> acceleration. And maybe it isn't all that useful because except for a
> minority of keyboard lovers (who really want single-key accelerators)
> most users will anyway use the mouse to select something in a menu.
I think this is an interesting discussion about flattening heirarchies of menus 
in both the system menu and application menus:

http://brianwill.net/blog/2009/08/03/reinventing-the-desktop-part-2-i-heard-
you-like-lists-now-in-text-form/

If each shortcut character sequence is matching against the whole menu 
heirarchy I would have thought that often a match would involve less key 
strokes than the 'alt+this', 'alt+that' style.

-- Richard