Re: non-latin accelerator keys

[Matthias Clasen <mclasen redhat com>]

On Thu, 2005-12-22 at 03:47 +0800, Abel Cheung wrote:
> Hi,
> 
> On 12/22/05, Matthias Clasen <mclasen redhat com> wrote:
> > http://bugzilla.gnome.org/show_bug.cgi?id=323956
> > http://bugzilla.gnome.org/show_bug.cgi?id=104112
> >
> > The first bug complains about the fact that the "(_F)" form in which
> > many CJK strings display the accelerator is not fully stripped out
> > when showing the string in a toolbar, and you end up with "(F)" in
> > the visible string.
> >
> > I am considering to change gtk_toolbar_elide_underscores() to strip not
> > only lone _ characters, but also a sequence of the form " (_<single
> > character>)" at the end of the string.
> >
> > I have a number of questions here:
> > - Does this sound like a reasonable thing to do ? (the risk of
> >   accidentally stripping something thats not an accelerator is
> >   probably minimal, but not 0.
> 
> Indeed, there can be cases where a single CJK character is enclosed inside
> parenthesis, and that's not uncommon; although when enclosed
> character is a latin character it mostly means mnemonic key.

But in the case of a single CJK character in parenthesis, it will likely
not have an _ before it, right ?

> 
> > - Is the (_F) approach generally considered just a workaround for
> >   the second bug, or are there languages where it is the
> >   preferred/standard way to display accel keys ?
> 
> Well, it is preferred, since multiple keystrokes are needed to input
> non-latin characters, and I doubt if anything like Alt-<char> can
> be entered at all. Hope anybody can enlighten me if this is
> possible or not.
> 

No, as you said, you normally need multiple keystrokes, unless you use
an exotic keymap.

>  The other bug asks for a way to underline a character in the label,
> > but have a different character as accel key.  I wrote patches which
> > change the Pango/GTK+ behaviour in the following way:
> >
> > f_oo  -> o underlined, accel key o
> > f_[x]oo  -> o underlined, accel key x
> >
> > Essentially the same questions here:
> > - Does this sound like a reasonable thing to do ? (the risk of
> >   accidentally stripping something thats not an accelerator is
> >   probably minimal, but not 0.
> 
> One question not entirely related: let's say "f_[x]oo", is there
> any hint or visual indication that the accel key is "x" not "o"?

No, but the use case for this is probably situations where the user
understand that the underlined 'o' means that he has to press the x
key. E.g. if there is a standard input method which maps the x key
to the 'o' character.

Matthias